E111: Microsoft to invest $10B in OpenAI, generative AI hype, America's over-classification problem

00:00:00.000 | Is anybody else seeing a half a second lag with J. Cole?

00:00:03.360 | Like a second lag?

00:00:04.440 | Test, test, test.

00:00:05.080 | One, two, one, two.

00:00:05.840 | From like the way his mouth moves.

00:00:07.720 | Well, that always happens.

00:00:09.200 | Oh, God.

00:00:09.720 | Here it comes.

00:00:10.440 | His mouth never stops moving.

00:00:12.080 | Relax, Sax.

00:00:13.160 | Relax, Sax.

00:00:16.040 | Are we going?

00:00:16.640 | Are we recording?

00:00:17.600 | Are you ready to go?

00:00:18.600 | Don't lose this.

00:00:19.720 | Don't lose this.

00:00:20.360 | A plus material.

00:00:21.400 | Are we going to do hot sacks?

00:00:22.360 | All right, let's go.

00:00:23.000 | This is Chappelle at the punchline.

00:00:24.640 | Let's go.

00:00:25.120 | Let's go.

00:00:25.640 | I'm ready to go.

00:00:26.440 | [MUSIC PLAYING]

00:00:28.240 | Let your winners ride.

00:00:30.960 | Rain Man David Saxon.

00:00:32.240 | I'm going all in.

00:00:35.200 | And it's said we open sourced it to the fans,

00:00:37.200 | and they've just gone crazy with it.

00:00:39.160 | Love you guys.

00:00:39.660 | Queen of Kinwam.

00:00:41.160 | I'm going all in.

00:00:43.160 | All right, everybody.

00:00:44.160 | Welcome to episode 111 of the World's Greatest Podcasts,

00:00:50.720 | according to Slate, the podcast that shall not be mentioned

00:00:54.320 | by the press, apparently.

00:00:55.680 | No, what do you mean?

00:00:56.140 | They just did a profile on us.

00:00:57.600 | Well, they did.

00:00:58.480 | This is the conundrum.

00:01:01.120 | It's so much of a phenomenon that we're

00:01:03.040 | the number one business and the number one tech

00:01:05.120 | podcast in the world, hands down,

00:01:07.840 | that the press has a hard time giving us any oxygen,

00:01:13.240 | because they want to hate us.

00:01:16.760 | They want to cover it.

00:01:17.840 | You're saying they take the ideas, but not the--

00:01:20.360 | They don't want to cite it.

00:01:21.600 | They don't want to cite it.

00:01:22.680 | They don't want to cite it.

00:01:23.520 | But anyway, shout out to Slate.

00:01:25.120 | Yeah, what I thought was interesting

00:01:26.080 | was the guy pointed out that we don't

00:01:28.560 | want to subject ourselves to independent journalists asking

00:01:31.440 | us independent questions.

00:01:32.960 | Therefore, we go direct.

00:01:34.160 | And that's kind of the thing nowadays.

00:01:36.760 | When everyone says they want to go direct,

00:01:38.560 | it's because they don't want to be subject

00:01:39.680 | to independent journalists.

00:01:40.840 | Well, one might ask themselves why subjects

00:01:43.960 | don't want to go direct.

00:01:44.920 | Yeah, exactly.

00:01:45.880 | You mean don't want to go to journalists.

00:01:47.640 | Yeah, because there's a specific reason why

00:01:50.360 | principals, the subject of stories,

00:01:52.160 | do not want to have the press interpret what they're

00:01:55.800 | saying is because they don't feel

00:01:56.680 | they're getting a fair shake.

00:01:57.640 | They feel like the world should be interesting.

00:01:59.000 | The challenge is that then we avoid

00:02:00.520 | independent scrutiny of our points of view

00:02:03.160 | and our decisions.

00:02:04.160 | They're constantly writing hit pieces about us.

00:02:06.440 | The question is, when we want to present our side of it,

00:02:10.680 | do we need to go through their filter or not?

00:02:12.760 | Why would you go through their filter when it's always

00:02:15.120 | going to be a hit piece?

00:02:17.360 | Well, and also--

00:02:18.200 | They have a class hatred of basically of technology

00:02:23.400 | entrepreneurs and investors.

00:02:24.640 | Just use sex.

00:02:25.560 | By the way, I think--

00:02:26.440 | No, I don't think--

00:02:27.240 | Just use sex.

00:02:27.960 | You're right, J-Kal.

00:02:28.800 | They don't hate you because you genuflect

00:02:30.640 | to their political biases.

00:02:31.760 | You see, if--

00:02:33.080 | Hold on.

00:02:34.720 | If you do what SPF did, which is basically

00:02:38.640 | agree with all of their biases, then yes,

00:02:41.600 | they'll treat you better.

00:02:42.600 | That's the deal.

00:02:43.560 | That's how it works.

00:02:44.560 | And when you say they, you're referring

00:02:46.040 | to specific large media outlets, right, Saks?

00:02:48.240 | They all think the same way.

00:02:49.440 | He's not referring to Fox and Tucker.

00:02:52.640 | OK, you can name one.

00:02:53.880 | I'll trade you.

00:02:54.600 | I'll tell you what.

00:02:55.400 | I'll trade you Fox for MSNBC and CNN and The New York Times,

00:02:59.000 | The Washington Post, and The Atlantic Magazine,

00:03:01.000 | and on and on and on.

00:03:02.200 | You get a lot of mileage out of being able to name Fox.

00:03:04.480 | The fact of the matter is--

00:03:05.560 | Megyn Kelly?

00:03:07.480 | That's a podcaster.

00:03:08.240 | She's independent now.

00:03:09.160 | That's true.

00:03:09.680 | She is independent.

00:03:10.480 | You can name one, I mean, literally one outlet that

00:03:14.160 | is not part of this mainstream media.

00:03:17.480 | And they all think the same way.

00:03:19.480 | There are very small differences in the way they think.

00:03:21.720 | It's all about clicks.

00:03:22.920 | It's all about clicks at this point.

00:03:24.880 | And it's all about advocacy journalism.

00:03:26.560 | Not just about clicks.

00:03:27.480 | And advocacy.

00:03:28.040 | It's that combination.

00:03:29.080 | What you're calling advocacy is bias and activism.

00:03:32.760 | It's activism.

00:03:33.920 | That's what I'm talking about, activism journalism, yes.

00:03:36.200 | I think Draymond also highlights a really important point,

00:03:38.840 | which is he started his podcast.

00:03:40.920 | It's become one of the most popular forms of sports media.

00:03:45.240 | And he can speak directly without the filtering

00:03:47.640 | and classification that's done by the journalist.

00:03:52.840 | And it seems to be a really powerful trend.

00:03:54.760 | The audience really wants to hear direct

00:03:56.480 | and they want to hear unfiltered, raw points of view.

00:04:00.640 | And maybe there's still a role for, I think,

00:04:03.040 | the journalism separate from that, which

00:04:05.680 | is to then scrutinize and analyze and question and--

00:04:08.360 | It's not journalism.

00:04:09.320 | It's just activism.

00:04:10.480 | They're just activists.

00:04:13.080 | There are also journalists out there, Sax.

00:04:15.000 | Actually, well, it depends what the topic is

00:04:17.840 | and what the outlet is.

00:04:19.680 | But actually, I would argue that most of these journalists

00:04:23.040 | are doing what they're doing for the same reason

00:04:25.080 | that we're doing what we're doing,

00:04:26.480 | which is they want to have some kind of influence

00:04:29.080 | because they don't get paid very much.

00:04:31.200 | But the way they have influence is

00:04:32.720 | to push a specific political agenda.

00:04:34.720 | I mean, they're activists.

00:04:36.080 | They're basically party activists.

00:04:37.480 | It has become advocacy journalism.

00:04:38.680 | Yes, that's the term I coined for it.

00:04:40.120 | It's advocacy journalism.

00:04:41.160 | You guys see this brouhaha where Matt Iglesias

00:04:44.440 | wrote this article about the Fed and about the debt ceiling?

00:04:49.120 | And through this whole multi-hundred word,

00:04:51.960 | thousand word tome, he didn't understand

00:04:57.400 | the difference between a percentage point and a basis

00:04:59.440 | point and then he didn't calculate the interest

00:05:01.280 | correctly?

00:05:02.000 | Yeah, I did see that.

00:05:04.400 | Wow, so wait a second.

00:05:05.680 | You're saying the Fed's raising 25%?

00:05:08.760 | Yeah, that's a huge difference between--

00:05:11.520 | My mortgage is going up 25%?

00:05:13.520 | Between a principal and an outside analyst, right?

00:05:15.720 | Like a principal has a better grasp typically

00:05:18.400 | of the topics and the material.

00:05:19.840 | But the argument from a journalist--

00:05:21.960 | But he's considered, within the journalist circle,

00:05:25.640 | he's considered the conventional wisdom.

00:05:27.320 | I get it.

00:05:27.800 | But the argument from a journalist

00:05:29.200 | is that by having that direct access,

00:05:31.880 | that person is also biased.

00:05:33.360 | Because they're an agent, because they're

00:05:35.120 | a player on the field, they do have a point of view

00:05:37.120 | and they do have a direction they want to take things.

00:05:39.320 | So it is a fair commentary that journalists can theoretically

00:05:42.880 | play a role, which is they're an off-field analyst

00:05:45.480 | and don't necessarily bring bias.

00:05:47.200 | I would argue they're less educated and more biased

00:05:50.080 | than we are.

00:05:50.680 | That may or may not be true, what the two of you

00:05:52.480 | guys are debating, which is a very subjective take.

00:05:55.040 | But the thing that is categorical and you can't deny

00:05:57.480 | is that there is zero checks and balances when something

00:06:02.120 | as simple as the basis point, percentage point difference

00:06:05.520 | isn't caught in proofreading, isn't caught by any editor,

00:06:09.360 | isn't caught by the people that help them review this.

00:06:12.120 | And so what that says is all kinds of trash

00:06:15.040 | must get through because there's no way for the average person

00:06:18.920 | on Twitter to police all of this nonsensical content.

00:06:22.200 | This one was easy because it was so numerically illiterate

00:06:25.280 | that it just stood out.

00:06:27.000 | But can you imagine the number of unforced errors journalists

00:06:31.920 | make today in their search for clicks that don't get caught

00:06:35.320 | out, that may actually tip somebody to think A versus B?

00:06:39.040 | That's, I think, the thing that's kind of undeniable.

00:06:42.120 | Right.

00:06:42.880 | Yeah.

00:06:43.400 | You only need to--

00:06:44.200 | Wasn't the Rolling Stone article?

00:06:44.800 | There's a very simple test for this.

00:06:46.760 | If you read the journalists writing about a topic

00:06:50.720 | you are an expert on, whatever the topic happens to be,

00:06:54.040 | you start to understand, OK, well,

00:06:55.760 | on that story I'm reading, that they understand about 10%

00:07:00.160 | or 20% or 30% of what's going on.

00:07:03.280 | But then when you read stories that you're not involved in,

00:07:05.740 | you know, you read a story about Hollywood or, I don't know,

00:07:08.320 | pick an industry or a region you're not super aware of,

00:07:11.160 | you're like, OK, well, that must be 100% correct.

00:07:14.040 | And the truth is, journalists have access to 5 to 20--

00:07:16.400 | Yeah, there's a name for that.

00:07:17.160 | There is a name for it, yeah.

00:07:18.080 | It's called the "Gentleman Amnesia Effect."

00:07:19.680 | You just plagiarized Michael Crichton,

00:07:21.360 | who came up with that.

00:07:22.560 | Yeah, so you--

00:07:23.800 | Yeah.

00:07:24.560 | But no, he's exactly right.

00:07:25.800 | But I think it's worse than that.

00:07:27.180 | It's because now the mistakes aren't being driven just

00:07:31.120 | by sloppiness or laziness or just a lack of expertise.

00:07:34.640 | I think it's being driven by an agenda.

00:07:37.080 | So just to give you an example on the Slate thing,

00:07:39.320 | the Slate article actually wasn't bad.

00:07:40.940 | It kind of made us seem, you know, cool.

00:07:43.360 | The subheadline was, "A close listen

00:07:45.520 | to all in the infuriating, fascinating, safe space

00:07:48.400 | for Silicon Valley's money men."

00:07:49.720 | OK.

00:07:50.560 | But the headline changed.

00:07:52.380 | So I don't know if you guys noticed this.

00:07:54.080 | The headline now is, "Elon Musk's Inner Circle

00:07:57.440 | is telling us exactly what it thinks."

00:07:59.140 | First of all, like, they're trying to--

00:08:00.840 | It's Elon for clicks.

00:08:01.640 | Yeah, it's Elon for-- so they're trying way too hard to,

00:08:03.900 | like, describe us in terms of Elon, which, you know,

00:08:07.000 | is maybe two episodes out of 110.

00:08:09.280 | But before Inner Circle, the word they used was cronies.

00:08:12.960 | And then somebody edited it, because I

00:08:14.560 | saw cronies in, like, one of those tweet, you know,

00:08:19.680 | summaries.

00:08:21.440 | You know, where, like, it does a capsule or whatever?

00:08:23.960 | Yeah, yeah, yeah.

00:08:24.680 | And those get frozen in time.

00:08:26.520 | So, you know, they were trying to bash us even harder.

00:08:28.920 | And then somebody took another look at it and toned it down.

00:08:30.960 | Well, here's what happened.

00:08:32.080 | I'll tell you what happens in the editorial process.

00:08:34.040 | Whoever writes the article, the article gets submitted.

00:08:36.580 | Maybe it gets edited, proofread, whatever.

00:08:38.800 | Maybe it doesn't even in some publications.

00:08:40.640 | They don't have the time for it, because they're in a race.

00:08:43.200 | Then they pick-- there's somebody who's

00:08:44.880 | really good at social media.

00:08:46.080 | They pick six or seven headlines.

00:08:47.640 | They A/B test them.

00:08:49.080 | And they even have software for this,

00:08:51.000 | where they will run a test.

00:08:52.360 | Sometimes they'll do a paid test.

00:08:53.760 | They put $5 in ads on social media.

00:08:57.400 | Whichever one performs the best, that's the one they go with.

00:09:00.520 | So it's even more cynical.

00:09:01.920 | And because people who read the headlines,

00:09:04.720 | sometimes they don't read the story, right?

00:09:06.480 | Obviously, most people just see the headline.

00:09:07.760 | They interpret that as a story.

00:09:09.280 | That's why I told you, when they did that New Republic

00:09:11.480 | piece on you with that horrific monstrosity of an illustration,

00:09:17.360 | don't worry about it.

00:09:18.280 | People just read the headline.

00:09:19.480 | They know you're important.

00:09:20.640 | Nobody reads the story anyway.

00:09:22.720 | But it wasn't a bad article, actually.

00:09:24.720 | It was well-written, actually.

00:09:25.760 | I was in shock.

00:09:26.480 | I was like, who is this writer that actually

00:09:28.320 | took the time to write some prose that was actually decent?

00:09:31.280 | Yeah, he had listened to a lot of episodes, clearly.

00:09:33.600 | That was a really good moment, actually.

00:09:34.880 | That was great advice, because you gave it to him.

00:09:36.640 | And you gave it to me, because both of us had these things.

00:09:39.100 | And Jason said the same thing.

00:09:40.880 | Just look at the picture.

00:09:42.320 | And if you're OK with the picture, just move on.

00:09:44.320 | And I thought, this can't be true.

00:09:46.200 | And it turned out to mostly be true.

00:09:47.920 | Yeah, but my picture was terrible.

00:09:50.400 | Yeah, but it's close to reality.

00:09:51.840 | So I mean, you've got to--

00:09:54.480 | It was really--

00:09:55.640 | Oh, ha, geez.

00:09:57.920 | I mean, the person who did the worst there was Peter Thiel.

00:10:00.880 | Poor Peter.

00:10:01.400 | Yeah, but that just shows how ridiculously biased it is,

00:10:03.800 | right?

00:10:05.480 | My picture wasn't so bad.

00:10:06.480 | My picture wasn't so bad.

00:10:08.200 | Hugh Grant.

00:10:09.100 | Elon, let's pull that up one more time here.

00:10:11.100 | Elon looks like Hugh Grant.

00:10:12.220 | I just kind of--

00:10:12.940 | Yeah, he does.

00:10:13.940 | Not bad.

00:10:14.900 | Kind of looks like Hugh Grant in "Notting Hill."

00:10:17.500 | I knew that article was going to be fine when

00:10:20.060 | the first item they presented as evidence of me doing something

00:10:24.700 | wrong was basically helping to oust Chasey Boudin, which

00:10:27.980 | was something that was supported by like 70% of San Francisco,

00:10:32.380 | which is a 90% Democratic city.

00:10:34.140 | So not exactly evidence of some out-of-control right-wing

00:10:38.060 | movement.

00:10:38.780 | Look at the headline.

00:10:40.420 | "The Quiet Political Rise of David Sachs, Silicon Valley's

00:10:43.500 | Prophet of Urban Doom."

00:10:44.780 | I'm just letting you know, people don't get past the six

00:10:46.940 | word in the image.

00:10:48.180 | It's 99% of people are like, oh my god,

00:10:50.620 | congrats on the "Republic" article.

00:10:52.660 | It could have literally been Laurel--

00:10:54.140 | what do they call them?

00:10:55.500 | Laurel Ipsums?

00:10:56.260 | You know, like it could have just

00:10:57.580 | been filler words from their second graph down

00:10:59.540 | and nobody would know.

00:11:00.900 | Yeah.

00:11:01.940 | But now apparently if you notice that San Francisco streets

00:11:06.020 | look like Walking Dead, that apparently you're

00:11:09.100 | a prophet of urban doom.

00:11:10.900 | I mean, these people are so out of touch.

00:11:12.580 | I mean, they can't even acknowledge what people

00:11:14.580 | can see with their own eyes.

00:11:16.020 | That's the bias that's gotten crazy.

00:11:17.540 | And I don't know if you guys saw this really horrible,

00:11:21.820 | dystopian video of an art gallery

00:11:24.140 | owner who's been dealing with owning a storefront in San

00:11:27.500 | Francisco, which is challenging, and having to clean up feces

00:11:31.860 | and trash and whatever every day.

00:11:36.020 | And I guess the guy snapped, and he's

00:11:38.180 | hosing down a homeless person who refuses to leave

00:11:42.020 | the front of his store.

00:11:43.700 | Oh, I saw that.

00:11:44.700 | I saw that.

00:11:45.500 | The humanity in this is just insane.

00:11:48.500 | Like, really, like you're hosing a human being down--

00:11:53.020 | It's terrible.

00:11:53.620 | --who is obviously not living a great life and is in dire--

00:11:57.980 | I can feel for both of them, J-Kal.

00:11:59.660 | I can feel for both of them.

00:12:00.940 | I agree that it's not good to hose a human being down.

00:12:03.740 | On the other hand, think about the sense of frustration

00:12:06.060 | that store owner has, because he's watching his business go

00:12:09.180 | in the toilet, because he's got homeless people living

00:12:11.780 | in front of him.

00:12:12.660 | So they're both, like, being mistreated.

00:12:15.220 | The homeless person's being mistreated.

00:12:16.900 | Say more about the store owner.

00:12:18.700 | Say more.

00:12:19.180 | The homeless person's being mistreated.

00:12:20.780 | The store owner's being mistreated

00:12:21.900 | by the city of San Francisco.

00:12:23.620 | Yeah.

00:12:24.820 | Say more about the store owner.

00:12:25.620 | That person's not in a privileged position.

00:12:27.660 | That person probably-- the store owner.

00:12:29.580 | He's probably fighting to stay in business.

00:12:32.100 | I'm just saying-- I'm not saying that's right, but--

00:12:34.180 | No, no.

00:12:34.660 | I'm just-- I'm laying the rope.

00:12:35.940 | No.

00:12:36.460 | I'm just--

00:12:36.940 | [INTERPOSING VOICES]

00:12:37.940 | --I'm just saying--

00:12:38.940 | [INTERPOSING VOICES]

00:12:39.940 | --justified in any way.

00:12:40.940 | What you're trying to do is, oh, my god,

00:12:43.260 | look at this homeless person being horribly oppressed.

00:12:45.700 | No.

00:12:46.180 | That store owner is a victim, too.

00:12:48.420 | Yeah, there's no doubt.

00:12:49.380 | It's horrible to run a business in there.

00:12:51.380 | I mean, what is that person supposed to do?

00:12:53.220 | No, wait.

00:12:53.740 | This is symbolic of the breaking down of basic society.

00:12:57.700 | Like, both of these people are obviously like--

00:13:01.780 | it's just a horrible moment to even witness.

00:13:04.060 | It's like, ugh.

00:13:06.020 | It's like something--

00:13:07.140 | Jason, do you have equal empathy for the store owner

00:13:10.180 | and the homeless person, or no?

00:13:12.700 | Under no circumstances should you

00:13:14.780 | hose a person down in the face who is homeless.

00:13:17.100 | Like, it's just horrific to watch.

00:13:18.620 | It's just inhumane.

00:13:19.700 | This is a human being.

00:13:20.980 | Now, but as a person who owns a store,

00:13:22.780 | yeah, my dad grew up in the local business.

00:13:24.540 | If people were abusing the store,

00:13:25.900 | and you're trying to make a living,

00:13:27.460 | and you've got to clean up whatever, excrement every day,

00:13:31.780 | which is horrific.

00:13:33.100 | Yes.

00:13:33.620 | And this thing is dystopian.

00:13:34.820 | In that moment, look, in that moment,

00:13:36.460 | the empathy is not equal.

00:13:37.740 | I think you have more empathy, obviously,

00:13:39.660 | for the person on the receiving end of that hose.

00:13:42.540 | But in general, our society has tons of empathy

00:13:47.260 | for homeless people.

00:13:48.100 | We spend billions of dollars trying to solve that problem.

00:13:50.740 | You never hear a thing about the store owners

00:13:53.340 | who are going out of business.

00:13:54.700 | So on a societal level, not in that moment, but in general,

00:13:59.740 | the lack of empathy is for these middle class store owners, who

00:14:03.740 | may not even be middle class, working class, who

00:14:06.260 | are struggling to stay afloat.

00:14:08.260 | And you look at something like, what is it,

00:14:10.220 | like a quarter or a third of the storefronts in San Francisco

00:14:13.340 | are now vacant?

00:14:14.420 | I'm just shocked this person is running--

00:14:16.980 | the shocking thing is, like, this person

00:14:18.580 | is running an art gallery storefront in San Francisco.

00:14:21.540 | Like, why would you even bother?

00:14:23.180 | Why would you bother to have a storefront in San Francisco?

00:14:25.620 | I mean, everybody's left.

00:14:26.740 | It's just--

00:14:27.260 | What do you mean, why do you bother?

00:14:28.780 | If you've opened a store, what are you supposed to do,

00:14:31.100 | start to code all of a sudden?

00:14:32.500 | Well, no.

00:14:33.020 | I mean, you would shut it down at some point and find an exit.

00:14:35.620 | And do what?

00:14:36.140 | Just like all businesses have.

00:14:37.340 | Yeah, but a store has large fixed costs, right?

00:14:38.940 | So what are they supposed to do?

00:14:40.220 | Maybe an investment he made 10 years ago.

00:14:41.940 | Exactly.

00:14:42.460 | At some point, you have to shut down your store

00:14:44.420 | in San Francisco the second you can get out of the lease.

00:14:46.220 | The solution to everything, J. Cal,

00:14:47.660 | isn't go to coding school online and then, you know,

00:14:49.900 | end up working for Google.

00:14:51.020 | Oh, I didn't say it was.

00:14:52.020 | Moving to another city is a possibility.

00:14:53.740 | So true.

00:14:54.820 | A lot of folks in Silicon Valley,

00:14:56.220 | I think, in this weirdly fucked up way,

00:14:58.140 | do believe the solution to everything is learn to code.

00:15:00.980 | Or become an Uber driver.

00:15:02.180 | Or become a masseuse.

00:15:03.980 | Get a gig job.

00:15:05.460 | Get a gig job.

00:15:07.540 | The guy spent years building his retail business.

00:15:09.940 | I mean, the thing is--

00:15:10.820 | And then a homeless person camps in front.

00:15:12.540 | And he calls the police.

00:15:13.660 | The police don't come and move the homeless person.

00:15:15.700 | The homeless person stays there.

00:15:16.700 | He asks nicely to move.

00:15:17.540 | Customers are uncomfortable going in the store as a result.

00:15:19.860 | Yeah, I stopped going to certain stores in my neighborhood

00:15:22.140 | because of homeless tents being literally fixated

00:15:24.780 | in front of the store.

00:15:25.380 | And I'd go to the store down the road

00:15:26.980 | to get my groceries or whatever.

00:15:28.300 | I mean, it's not a kind of uncommon situation

00:15:30.660 | for a lot of these small business owners.

00:15:32.380 | They don't own the real estate.

00:15:33.700 | They're paying rent.

00:15:35.140 | They've got high labor costs.

00:15:36.780 | Everything's inflating.

00:15:38.220 | Generally, city population's declining.

00:15:40.020 | It's a brutal situation all around.

00:15:43.540 | I think if everybody learns to code or drives an Uber,

00:15:46.860 | the problem is that in the absence of things

00:15:49.220 | like local stores and small businesses,

00:15:51.740 | you hollow out communities.

00:15:53.980 | You have these random detached places where you kind of live.

00:15:57.260 | And then you sit in your house, which

00:15:58.820 | becomes a prison while you order food from an app every day.

00:16:02.300 | I don't think that is the society that people want.

00:16:04.660 | So I don't know.

00:16:05.380 | I kind of want small businesses to exist.

00:16:08.180 | And I think that the homeless person should be taken care of.

00:16:10.780 | But the small business person should

00:16:12.380 | have the best chance of trying to be successful

00:16:14.340 | because it's hard enough as it is.

00:16:15.980 | The mortality rate of the small business owner

00:16:19.260 | is already 90%.

00:16:22.020 | It's impossible in San Francisco, let's just be honest.

00:16:24.260 | So stop genuflecting, J. Cal.

00:16:26.100 | I'm not genuflecting.

00:16:27.420 | Listen--

00:16:27.940 | Yeah, you are because here's--

00:16:29.460 | Wait, how am I genuflecting?

00:16:30.620 | I'm saying the guy--

00:16:31.500 | I'm just shocked that the guy even has a storefront.

00:16:33.500 | I would have left a long time ago.

00:16:34.380 | You're showing a tweet that's a moment in time.

00:16:36.460 | And you're not showing the 10 steps that led up to it.

00:16:39.620 | Oh, 1,000 steps.

00:16:40.660 | The five times he called the police to do something about it.

00:16:43.300 | I framed it as dystopian from the get.

00:16:44.940 | I'm not genuflecting.

00:16:45.980 | I'm not genuflecting.

00:16:46.980 | I'm not genuflecting.

00:16:47.900 | I'm not genuflecting.

00:16:48.820 | I'm not genuflecting.

00:16:49.740 | I'm not genuflecting.

00:16:50.660 | I'm not genuflecting.

00:16:51.620 | I'm not genuflecting.

00:16:52.540 | I'm not genuflecting.

00:16:53.460 | I'm not genuflecting.

00:16:54.380 | I'm not genuflecting.

00:16:55.260 | I'm not genuflecting.

00:16:56.100 | I'm not genuflecting.

00:16:56.980 | I'm not genuflecting.

00:16:57.860 | I'm not genuflecting.

00:16:58.740 | I'm not genuflecting.

00:16:59.620 | I'm not genuflecting.

00:17:00.500 | I'm not genuflecting.

00:17:01.420 | I'm not genuflecting.

00:17:02.340 | I'm not genuflecting.

00:17:03.220 | I'm not genuflecting.

00:17:04.140 | I'm not genuflecting.

00:17:05.020 | It's addiction.

00:17:05.740 | It's addiction.

00:17:06.700 | You say you know this, and it's mental illness.

00:17:08.620 | Schellenberger's done the work.

00:17:09.940 | It's like he said, 99% of the people he talks to,

00:17:12.740 | it's either mental illness or addiction.

00:17:14.500 | But we keep using this word homeless

00:17:16.500 | to describe the problem.

00:17:18.140 | But the issue here is not the lack of housing,

00:17:20.900 | although that's a separate problem in California.

00:17:22.980 | But it's basically the lack of treatment.

00:17:25.860 | Totally.

00:17:26.340 | So we should be calling them treatmentless.

00:17:27.860 | And mandates around this, because--

00:17:29.340 | And enforcement.

00:17:30.140 | You cannot have-- you can't have a super drug

00:17:33.540 | be available for a nominal price and give people

00:17:38.380 | a bunch of money to come here and take it and not enforce it.

00:17:40.980 | You have to draw the line at fentanyl.

00:17:43.380 | I'm sorry.

00:17:44.020 | Fentanyl is a super drug.

00:17:45.500 | There's three alternatives.

00:17:46.820 | There's mandated rehab, mandated mental health or jail,

00:17:52.140 | or housing services.

00:17:54.340 | If you're not breaking the law, you don't have mental illness,

00:17:56.860 | you don't have drug addiction.

00:17:57.700 | And then provide-- those are the four paths of outcome here

00:18:00.180 | of success.

00:18:00.900 | And if all four of those paths were both mandated

00:18:03.780 | and available in abundance, this could be a tractable problem.

00:18:08.500 | Unfortunately, the mandate-- I mean,

00:18:10.860 | you guys remember that Kevin Bacon movie,

00:18:12.820 | where Kevin Bacon was locked up in a mental institution,

00:18:16.220 | but he wasn't like--

00:18:18.100 | he wasn't mentally ill?

00:18:19.340 | It's a famous story.

00:18:20.420 | Footloose?

00:18:21.540 | What's that?

00:18:22.380 | Footloose.

00:18:22.900 | No.

00:18:23.380 | No, it's a famous story.

00:18:24.820 | You guys-- someone's probably going to call me an idiot

00:18:27.060 | for messing this whole thing up.

00:18:29.740 | But I think there's a story where mandated mental health

00:18:37.260 | services, like locking people up to take care of them

00:18:40.460 | when they have mental health issues like this,

00:18:43.260 | became kind of inhumane.

00:18:45.660 | And a lot of the institutions were shut down,

00:18:47.500 | and a lot of the laws were overturned.

00:18:49.300 | And there are many of these cases that happened,

00:18:51.620 | where they came across as torturous to what

00:18:53.860 | happened to people that weren't mentally ill.

00:18:55.980 | And so the idea was, let's just abandon the entire product.

00:18:58.500 | Like a goose nest?

00:18:59.780 | That's another good one.

00:19:00.820 | Yeah.

00:19:01.340 | Well, that's another one, right?

00:19:02.700 | And it's unfortunate, but I think that there's some--

00:19:05.060 | we talk a lot about nuance and gray areas,

00:19:06.940 | but there's certainly some solution here

00:19:09.020 | that isn't black or white.

00:19:10.100 | It's not about not having mandated mental health

00:19:12.020 | services, and it's not about locking everyone up

00:19:14.860 | that has some slight problem.

00:19:16.540 | But there's some solution here that needs to be crafted,

00:19:19.500 | where you don't let people suffer,

00:19:21.060 | and you don't let people suffer both as the victim

00:19:24.860 | on the street, but also the victim in the community.

00:19:27.020 | You're talking about a 51/50, I think,

00:19:28.020 | like when people are held because they're

00:19:30.660 | a danger to themselves or others kind of thing?

00:19:33.100 | Right, but Jacob, let's think about the power of language

00:19:34.900 | here.

00:19:35.400 | If we refer to these people as untreated persons instead

00:19:39.380 | of homeless persons, and that was the coverage 24/7

00:19:42.740 | in the media is, this is an untreated person,

00:19:45.220 | the whole policy prescription would be completely different.

00:19:47.720 | We'd realize there's a shortage of treatment.

00:19:50.020 | We'd realize there's a shortage of remedies related

00:19:52.220 | to getting people in treatment, as opposed to building housing.

00:19:56.740 | But why--

00:19:58.100 | And laws that mandate it, that don't enable it.

00:20:00.860 | Because if you don't mandate it, then you

00:20:02.780 | enable the free reign and the free living on the street

00:20:05.820 | and the open drug markets and all this sort of stuff.

00:20:08.140 | There's a really easy test for this.

00:20:10.460 | If it was yourself, and you were addicted,

00:20:14.180 | or if it was a loved one, one of your immediate family members,

00:20:16.980 | would you want yourself or somebody else

00:20:19.500 | to be picked up off the street and held with a 51/50

00:20:22.260 | or whatever code involuntarily against their will

00:20:26.220 | because they were a danger?

00:20:27.340 | Would you want them to be allowed to remain on the street?

00:20:29.740 | Would you want yourself if you were in that dire straits?

00:20:32.260 | And the answer, of course, is you

00:20:34.500 | would want somebody to intervene.

00:20:35.980 | But what's the liberal policy perspective on this, J. Cal?

00:20:38.320 | So let me ask you as our diehard liberal on the show--

00:20:41.060 | No, I'm not a diehard liberal.

00:20:42.100 | No.

00:20:42.460 | No, he's an independent and only votes for Democrats.

00:20:44.780 | Please, get it right.

00:20:45.660 | 75% of the time I vote a Democrat, 25% Republican.

00:20:47.940 | Please, get it right.

00:20:48.820 | Independent votes for Democrats.

00:20:50.460 | 25% Republicans.

00:20:51.900 | Is it not that your individual liberties are infringed upon

00:20:54.540 | if you were to be, quote, "picked up and put away"?

00:20:57.420 | My position on it is if you're not thinking straight

00:21:00.780 | and you're high on fentanyl, you're

00:21:02.700 | not thinking for yourself.

00:21:03.940 | And you could lose the liberty for a small period of time--

00:21:06.820 | 72 hours, a week--

00:21:09.340 | especially if you're a danger to somebody, yourself

00:21:12.620 | or other people.

00:21:13.420 | And in this case, if you're on fentanyl, if you're on meth,

00:21:15.940 | you're a danger to society.

00:21:17.740 | I think if more people had that point of view

00:21:21.020 | and had that debate, as Saxe is saying, in a more open way,

00:21:24.180 | you could get to some path to resolution on--

00:21:26.220 | Just not in San Francisco.

00:21:27.660 | It's not how it happened.

00:21:28.700 | So you guys know this.

00:21:30.940 | We won't say who it is, but someone in my family

00:21:34.500 | has some pretty severe mental health issues.

00:21:36.700 | And the problem is, because they're an adult,

00:21:41.660 | you can't get them to get any form of treatment whatsoever.

00:21:45.700 | Right, right.

00:21:47.660 | You only have the nuclear option.

00:21:49.740 | And the nuclear option is you basically take that person

00:21:52.020 | to court and try to seize their power of attorney, which

00:21:54.660 | is essentially saying that--

00:21:56.180 | Individual liberties are gone.

00:21:57.900 | And by the way, it is so unbelievably restrictive

00:22:00.940 | what happens if you lose that power of attorney

00:22:03.860 | and somebody else has it over you.

00:22:06.840 | It's just a huge burden that the legal system

00:22:10.660 | makes extremely difficult.

00:22:12.540 | And the problem--

00:22:13.340 | Well, some of the law is a backstop.

00:22:14.500 | If the person's committing something illegal,

00:22:16.460 | like camping out or doing fentanyl, meth, whatever,

00:22:20.180 | you can use the law as the backstop against personal

00:22:23.100 | liberty.

00:22:23.600 | You can't use the law.

00:22:23.900 | All that person can do is really get arrested.

00:22:25.820 | Even that is not a high enough bar

00:22:27.660 | to actually get power of attorney over somebody.

00:22:29.660 | The other thing that I just wanted you guys to know--

00:22:31.940 | I think you know this, but just a little historical context--

00:22:34.860 | is a lot of this crisis in mental health

00:22:37.580 | started because Reagan defunded all the psychiatric hospitals.

00:22:41.100 | He emptied them in California.

00:22:43.580 | And that compounded, because for whatever reason,

00:22:46.420 | his ideology was that these things should

00:22:48.780 | be treated in a different way.

00:22:50.540 | And when he got to the presidency,

00:22:52.700 | one of the things that he did was

00:22:54.100 | he repealed the Mental Health--

00:22:55.700 | I think it's called the Mental Health Systems Act, MHSA,

00:23:00.420 | which completely broke down some pretty landmark legislation

00:23:03.260 | on mental health.

00:23:04.860 | And I don't think we've ever really recovered.

00:23:06.900 | We're now 42 years onward from 1980, or 43 years onward.

00:23:11.420 | But just something for you guys to know that that's--

00:23:14.900 | Reagan had a lot of positives, but that's

00:23:16.860 | one definitely negative check in my book against his legacy

00:23:21.060 | is his stance on mental health in general

00:23:22.740 | and what he did to defund mental health.

00:23:24.540 | Well, let me make two points there.

00:23:26.740 | So I'm not defending that specific decision.

00:23:29.400 | There were a bunch of scandals in the 1970s,

00:23:32.220 | and epitomized by the movie One Flew

00:23:33.860 | Over the Cuckoo's Nest with Jack Nicholson,

00:23:35.860 | about the conditions in these mental health homes.

00:23:39.580 | And that did create a groundswell

00:23:41.780 | to change laws around that.

00:23:43.780 | But I think this idea that somehow Reagan

00:23:47.100 | is to blame when he hasn't been in office for 50 years,

00:23:50.060 | as opposed to the politicians who've

00:23:51.540 | been in office for the last 20 years,

00:23:53.540 | I just think it's letting them off the hook.

00:23:55.340 | I mean, Gavin Newsom, 10, 15 years ago

00:23:57.900 | when he was mayor of San Francisco,

00:23:59.460 | declared that he would end homelessness within 10 years.

00:24:03.240 | He just made another declaration like that as governor.

00:24:06.180 | So I just feel like--

00:24:07.500 | I'm not saying it's Reagan's fault.

00:24:09.460 | I'm just saying--

00:24:10.180 | Well, I just think it's letting these guys off the hook.

00:24:11.340 | --it's an interesting historical moment.

00:24:12.220 | I think it's letting the politicians off the hook.

00:24:14.380 | Society needs to start thinking about changing priorities.

00:24:17.260 | We didn't have this problem of massive numbers of people

00:24:20.340 | living on the streets 10, 15 years ago.

00:24:23.100 | It was a much smaller problem.

00:24:24.700 | And I think a lot of it has to do with fentanyl.

00:24:26.660 | The power of these drugs has increased massively.

00:24:29.660 | There's other things going on here.

00:24:31.540 | So in any event, I mean, you can question what Reagan did

00:24:35.780 | in light of current conditions.

00:24:37.060 | But I think this problem really started in the last 10,

00:24:39.820 | 15 years.

00:24:41.100 | Like in an order of magnitude bigger way.

00:24:43.580 | These are super drugs.

00:24:44.500 | Until people realize these are a different class of drugs,

00:24:47.500 | and they start treating them as such,

00:24:49.700 | it's going to just get worse.

00:24:50.860 | There's no path.

00:24:52.500 | Oh, as far as I know, Reagan didn't hand out

00:24:55.620 | to these addicts $800 a week to feed their addiction

00:24:59.020 | so they could live on the streets of San Francisco.

00:25:01.140 | That is the current policy of the city.

00:25:02.900 | I hear you.

00:25:03.400 | It's a terrible policy.

00:25:04.400 | All I just wanted to just provide

00:25:06.140 | was just that color that we had a system of funding

00:25:10.180 | for the mental health infrastructure,

00:25:11.700 | particularly local mental health infrastructure.

00:25:14.180 | And we took that back.

00:25:16.460 | And then we never came forward.

00:25:18.180 | And all I was saying is I'm just telling you

00:25:20.580 | where that decision was made.

00:25:21.780 | I think that's part of the solution here is, yeah,

00:25:23.860 | we're going to have to basically build up shelters.

00:25:26.380 | We're going to have to build up homes.

00:25:28.060 | And to support your point, the problem now, for example,

00:25:31.220 | is Gavin Newsom says a lot of these things.

00:25:34.180 | And now he's gone from a massive surplus

00:25:37.380 | to a $25 billion deficit overnight, which we talked

00:25:40.780 | about even a year ago because that was just

00:25:43.140 | the law of numbers catching up with the state of California.

00:25:46.940 | And he's not in a position now to do any of this stuff.

00:25:49.260 | So this homeless problem may get worse.

00:25:51.540 | Well, they did appropriate--

00:25:53.460 | I forget the number.

00:25:54.380 | It's like $10 billion or something out of that huge

00:25:56.620 | budget they had to solve the problem of homelessness.

00:25:58.460 | I would just argue they're not tackling it in the right way.

00:26:01.060 | Because what happened is there's a giant special interest

00:26:04.220 | that formed around this problem, which

00:26:06.340 | is the building industry, who gets these contracts to build

00:26:12.060 | the, quote, "affordable housing" or the--

00:26:14.460 | Homeless industrial complex is insane.

00:26:16.020 | The homeless industrial complex.

00:26:17.380 | And so they end up building 10 units at a time

00:26:20.420 | on Venice Beach, like the most expensive land you could

00:26:23.540 | possibly build because you get these contracts

00:26:26.140 | from the government.

00:26:27.060 | So there's now a giant special interest

00:26:29.020 | in lobby that's formed around this.

00:26:31.020 | If you really want to solve the problem,

00:26:33.100 | you wouldn't be building housing on Venice Beach.

00:26:36.940 | You'd be going to cheap land just outside the city.

00:26:40.020 | Totally.

00:26:40.660 | And you'd be building scale shelters.

00:26:42.820 | I mean, shelters that can house 10,000 people, not 10.

00:26:47.900 | And you'd be having treatment services--

00:26:49.660 | But they also have to get treatment.

00:26:49.820 | Yes.

00:26:50.140 | You have to get treatment.

00:26:50.780 | But with treatment built into them, right?

00:26:52.660 | You'd be solving this problem at scale.

00:26:54.740 | And that's not what they're doing.

00:26:56.860 | By the way, do you guys want to hear this week in Grift?

00:27:00.180 | Sure.

00:27:01.020 | We're all in.

00:27:01.820 | That's a great example of Grift.

00:27:03.260 | I read something today in Bloomberg that was unbelievable.

00:27:07.260 | There's about $2 trillion of debt

00:27:09.780 | owned by the developing world that

00:27:13.180 | has been classified by a nonprofit, the Nature

00:27:16.300 | Conservancy in this case, as eligible for what

00:27:18.980 | they called nature swaps.

00:27:20.940 | So this is $2 trillion of the umpteen trillions of debt

00:27:23.780 | that's about to get defaulted on by countries like Belize,

00:27:27.580 | Ecuador, Sri Lanka, Seychelles, you name it.

00:27:31.900 | And what happens now are the big bulge bracket,

00:27:35.220 | Wall Street banks and the Nature Conservancy,

00:27:38.700 | goes to these countries and says, listen,

00:27:41.660 | you have a billion dollar tranche of debt that's

00:27:44.140 | about to go upside down.

00:27:46.660 | And you're going to be in default with the IMF.

00:27:48.980 | We'll let you off the hook.

00:27:50.940 | And we will negotiate with those bondholders

00:27:54.340 | to give them $0.50 on the dollar.

00:27:56.900 | But in return, you have to promise

00:27:59.420 | to take some of that savings and protect the rainforest

00:28:03.260 | or protect a coral reef or protect some mangrove trees.

00:28:07.180 | All sounds good.

00:28:08.740 | Except then what these folks do is

00:28:10.540 | they take that repackaged debt.

00:28:13.380 | They call it ESG.

00:28:15.340 | They mark it back up.

00:28:16.580 | And then they sell it to folks like BlackRock

00:28:18.900 | who have decided that they must own this in the portfolio.

00:28:21.980 | So it literally just goes from one sleeve of BlackRock,

00:28:24.940 | which is now marked toxic emerging market debt.

00:28:29.100 | And then it gets into someone's 401(k) as ESG debt.

00:28:33.460 | Isn't that unbelievable?

00:28:34.660 | So you could virtue signal and buy some ESG

00:28:37.340 | to make yourself feel good, yeah.

00:28:38.820 | $2 trillion of this paper.

00:28:39.940 | Here's all you have to know about ESG

00:28:41.580 | is that Exxon is like the number seven top-ranked company

00:28:46.020 | according to ESG.

00:28:47.340 | And Tesla is even on the list.

00:28:48.740 | Disastrous.

00:28:50.060 | How crazy is that?

00:28:52.780 | It's a complete scam.

00:28:54.620 | Yeah.

00:28:55.420 | All of those-- we've said this many times,

00:28:57.180 | but each of those letters individually

00:28:58.820 | means so much and should be worth a lot to a lot of people.

00:29:02.220 | But when you stick them together,

00:29:03.620 | it creates this toxic soup where you can just hide the cheese.

00:29:07.540 | Yeah, I mean, governance is important in companies.

00:29:10.260 | Of course, the environment is important.

00:29:12.660 | Social change is important.

00:29:14.180 | I mean, but why are these things grouped together in this?

00:29:17.620 | It just perverts the whole thing.

00:29:18.980 | It's an industry, J. Cal.

00:29:19.460 | It's an industry of consultants.

00:29:20.820 | It's a grift.

00:29:21.300 | Absolutely.

00:29:22.020 | All right, speaking of grifts, Microsoft

00:29:24.500 | is going to put $10 billion or something into chat GPT.

00:29:28.340 | Degenerate AI, as I'm calling it now,

00:29:30.820 | is the hottest thing in Silicon Valley.

00:29:34.420 | The technology is incredible.

00:29:35.740 | I mean, you can question the business model, maybe,

00:29:38.100 | but the technology is pretty--

00:29:39.660 | Well, I mean, yeah.

00:29:40.460 | So what I'd say is $29 billion for a company that's

00:29:43.180 | losing $1 billion in Azure credits a year.

00:29:45.700 | That's one way to look at it.

00:29:47.220 | That's also a naive way to look at a lot of other businesses

00:29:50.380 | that ended up being worth a lot down the road.

00:29:52.300 | I mean--

00:29:52.900 | Sure.

00:29:53.820 | You can model out the future of a business like this

00:29:56.260 | and create a lot of really compelling big outcomes.

00:30:00.340 | Potentially, yeah.

00:30:01.780 | So Microsoft is close to investing $10 billion

00:30:04.060 | in open AI in a very convoluted transaction

00:30:06.620 | that people are trying to understand.

00:30:08.780 | It turns out that they might wind up owning 59% of open AI,

00:30:13.900 | but get 75% of the cash and profits back over time.

00:30:18.020 | 49%.

00:30:19.300 | 49%, yeah, of open AI.

00:30:21.220 | But they would get paid back the $10 billion

00:30:23.780 | over some amount of time.

00:30:26.140 | And this obviously includes Azure credits and chat GPT.

00:30:32.340 | As everybody knows, this just incredible demonstration

00:30:36.980 | of what AI can do in terms of text-based creation of content

00:30:41.500 | and answering queries has taken the net by storm.

00:30:45.060 | People are really inspired by it.

00:30:47.100 | Sax, do you think that this is a defensible, real technology?

00:30:51.820 | Or do you think this is like a crazy hype cycle?

00:30:55.540 | Well, it's definitely the next VC hype cycle.

00:30:57.660 | Everyone's kind of glomming onto this,

00:30:59.260 | because VC really right now needs a savior.

00:31:01.620 | Just look at the public markets, everything we're investing in,

00:31:04.220 | it's in the toilets.

00:31:05.060 | So we all really want to believe that this

00:31:07.260 | is going to be the next wave.

00:31:10.140 | And just because something is a VC hype cycle

00:31:13.060 | doesn't mean that it's not true.

00:31:14.980 | So as I think one of our friends pointed out,

00:31:19.180 | mobile turned out to be very real.

00:31:21.020 | I think cloud turned out to be, I'd say, very real.

00:31:25.500 | Social was sort of real in the sense

00:31:28.020 | that it did lead to a few big winners.

00:31:30.460 | On the other hand, web 3 and crypto was a hype cycle

00:31:34.860 | and it's turned into a big bust.

00:31:36.180 | VR falls into the hype cycle.

00:31:37.500 | I think VR is probably a hype cycle so far.

00:31:39.860 | No one can even explain what web 3 is.

00:31:42.780 | In terms of AI, I think that if I had to guess,

00:31:47.300 | I would say the hype is real in terms

00:31:50.180 | of its technological potential.

00:31:53.100 | However, I'm not sure about how much potential there is yet

00:31:57.340 | for VCs to participate.

00:31:58.780 | Because right now, it seems like this

00:32:01.580 | is something that's going to be done by really big companies.

00:32:04.620 | So open AI is basically a--

00:32:07.140 | it looks like kind of a Microsoft proxy.

00:32:09.460 | You've got Google, I'm sure, will develop it

00:32:11.340 | through their DeepMind asset.

00:32:14.460 | I'm sure Facebook is going to do something huge in AI.

00:32:18.100 | So what I don't know is, is this really

00:32:20.180 | a platform that startups are going to be able to benefit

00:32:22.940 | from?

00:32:23.700 | I will say that some of the companies we've invested in

00:32:26.740 | are starting to use these tools.

00:32:28.180 | So I guess where I am is I think the technology is actually

00:32:34.460 | exciting.

00:32:35.780 | I wouldn't go overboard on the valuations.

00:32:39.220 | I wouldn't buy into that level of the hype.

00:32:41.020 | But you think there could be hundreds

00:32:42.700 | of companies built around an API for something

00:32:46.780 | like ChatGBT, Dolly--

00:32:48.140 | Maybe, yeah.

00:32:48.740 | I don't think startups are going to be able to create

00:32:51.020 | the AI themselves.

00:32:53.500 | But they might be able to benefit from the APIs.

00:32:56.940 | Maybe that's the thing that has to be proven out.

00:32:59.580 | Freebird.

00:33:00.100 | There's a lot of really fantastic machine learning

00:33:04.020 | services available through cloud vendors today.

00:33:07.100 | So Azure has been one of these kind of vendors.

00:33:09.940 | And obviously, OpenAI is building tools a little bit

00:33:13.500 | further down on the stack.

00:33:16.180 | But there's a lot of tooling that

00:33:17.660 | can be used for specific vertical applications.

00:33:19.780 | Obviously, the acquisition of InstaDeep by BioNTech

00:33:22.900 | is a really solid example.

00:33:24.180 | And most of the big dollars that are flowing in biotech

00:33:26.780 | right now are flowing into machine learning applications

00:33:29.700 | where there's some vertical application of machine

00:33:31.740 | learning tooling and techniques around some specific problem

00:33:35.180 | set.

00:33:36.180 | And the problem set of mimicking human communication

00:33:40.540 | and doing generative media is a consumer application

00:33:45.300 | set that has a whole bunch of really interesting product

00:33:47.540 | opportunities.

00:33:48.620 | But let's not kind of be blind to the fact

00:33:51.140 | that nearly every other industry and nearly every other vertical

00:33:54.660 | is being transformed today.

00:33:55.980 | And there's active progress being made in funding

00:33:59.740 | and getting liquidity on companies and progress

00:34:02.980 | with actual products being driven by machine learning

00:34:05.540 | systems.

00:34:06.020 | There's a lot of great examples of this.

00:34:07.700 | So the fundamental capabilities of large data sets

00:34:11.900 | and then using these kind of learning techniques

00:34:15.380 | in software and statistical models

00:34:16.860 | to make kind of predictions and drive businesses forward

00:34:20.900 | in a way that they're not able to with just human knowledge

00:34:23.940 | and human capability alone is really real.

00:34:27.660 | And it's here today.

00:34:29.100 | And so I think let's not get caught up in the fact

00:34:31.180 | that there's this really interesting consumer market

00:34:33.660 | hype cycle going on, where these tools are not

00:34:36.060 | being kind of validated and generating

00:34:38.580 | real value across many other verticals and segments.

00:34:41.220 | Chamath, when you look at this Microsoft OpenAI deal,

00:34:44.700 | and you see something that's this convoluted,

00:34:47.220 | hard to understand, what does that signal to you

00:34:50.780 | as a capital allocator and company builder?

00:34:52.820 | I would put deals into two categories.

00:34:55.340 | One is easy and straightforward.

00:34:57.780 | And then two is, you know, cute by half or the two hard bucket.

00:35:02.300 | This is clearly in that second category,

00:35:05.100 | but it doesn't mean that--

00:35:05.940 | - Why is it in that category?

00:35:07.140 | - Well, it doesn't mean that it won't work.

00:35:08.740 | In our group chat with the rest of the guys,

00:35:11.660 | one person said, there's a lot of complex law

00:35:14.820 | when you go from a nonprofit to a for-profit.

00:35:19.360 | There's lots of complexity in deal construction.

00:35:22.260 | The original investors have certain things

00:35:24.660 | that they want to see.

00:35:26.580 | There may or may not be, you know, legal issues at play here

00:35:29.820 | that you encapsulated well in the last episode.

00:35:32.820 | I think there's a lot of stuff we don't know.

00:35:34.980 | So I think it's important to just like,

00:35:37.520 | give those folks the benefit of the doubt.

00:35:39.140 | But yeah, if you're asking me,

00:35:41.300 | it's in the two hard bucket for me to really take seriously.

00:35:43.800 | Now, that being said, it's not like I got shown the deal.

00:35:47.300 | So I can't comment.

00:35:49.020 | Here's what I will say.

00:35:50.980 | The first part of what Zach said,

00:35:52.260 | I think is really important for entrepreneurs

00:35:54.780 | to internalize, which is where can we make money?

00:35:59.420 | The reality is that, well, let me just take a prediction.

00:36:04.140 | I think that Google will open source their models

00:36:07.220 | because the most important thing that Google can do

00:36:11.420 | is reinforce the value of search.

00:36:14.460 | And the best way to do that is to scorch the earth

00:36:17.420 | with these models, which is to make them widely available

00:36:20.380 | and as free as possible.

00:36:22.160 | That will cause Microsoft to have to catch up.

00:36:25.740 | And that will cause Facebook to have to really look

00:36:28.180 | in the mirror and decide whether they're going to cap

00:36:32.460 | the betting that they've made on AR/VR

00:36:35.820 | and reallocate very aggressively to AI.

00:36:38.220 | I mentioned this in the, I did this Lex Friedman podcast,

00:36:41.380 | but that should be what Facebook does.

00:36:43.240 | And the reason is, if Facebook and Google and Microsoft

00:36:47.340 | have roughly the same capability and the same model,

00:36:50.640 | there's an element of machine learning

00:36:53.660 | that I think is very important,

00:36:54.900 | which is called reinforcement learning.

00:36:56.780 | And specifically it's reinforcement learning

00:36:58.700 | from human feedback, right?

00:37:00.220 | So these RLHF pipelines,

00:37:01.860 | these are the things that will make your stuff unique.

00:37:05.780 | So if you're a startup,

00:37:07.420 | you can build a reinforcement learning pipeline.

00:37:10.340 | How?

00:37:11.160 | You build a product that captures a bunch of usage.

00:37:13.460 | We talked about this before.

00:37:14.900 | That data set is unique to you as a company.

00:37:17.740 | You can feed that into these models,

00:37:19.740 | get back better answers, you can make money from it.

00:37:22.380 | Facebook has an enormous amount of reinforcement learning

00:37:26.180 | inside of Facebook.

00:37:27.140 | Every click, every comment, every like, every share.

00:37:32.100 | Twitter has that data set.

00:37:33.940 | Google inside of Gmail and search.

00:37:36.300 | Microsoft inside of Minecraft and Hotmail.

00:37:40.520 | So my point is, David's right.

00:37:42.700 | The huge companies, I think, will create the substrates.

00:37:47.700 | And I think there'll be forced to scorch the earth

00:37:52.060 | and give it away for free.

00:37:54.340 | And then on top of that is where you can make money.

00:37:56.820 | And I would just encourage entrepreneurs to think,

00:37:59.340 | where is my edge in creating a data set

00:38:02.540 | that I can use for reinforcement learning?

00:38:04.200 | That I think is interesting.

00:38:05.620 | That's kind of saying,

00:38:06.900 | I buy the ingredients from the supermarket,

00:38:09.780 | but then I can still construct a dish that's unique.

00:38:12.700 | And the salt is there, the pepper is there,

00:38:15.680 | but how I use that will determine

00:38:17.540 | whether you like the thing or not.

00:38:19.240 | And I think that that is the way

00:38:21.340 | that I think we need to start thinking about it.

00:38:23.380 | - Interestingly, as we've all pointed out here,

00:38:26.100 | OpenAI was started as a nonprofit.

00:38:28.780 | The stated philosophy was this technology is too powerful

00:38:31.620 | for any company to own.

00:38:33.660 | Therefore, we're going to make it open source.

00:38:35.900 | And then somewhere in the last couple of years,

00:38:37.300 | they said, well, you know what, actually,

00:38:38.820 | it's too powerful for it to be out there in the public.

00:38:42.020 | We need to make this a private company

00:38:44.380 | and we need to get $10 billion from Microsoft.

00:38:48.000 | That is the disconnect I am trying to understand.

00:38:51.700 | That's the most interesting part of the story, Jason,

00:38:54.060 | I think, if you go back to 2014,

00:38:56.460 | is when Google bought DeepMind.

00:38:58.540 | And immediately everyone started reacting

00:39:01.560 | to a company as powerful as Google,

00:39:03.540 | having a toolkit and a team as powerful

00:39:06.100 | as DeepMind within them.

00:39:07.940 | And that sort of power should not sit in anyone's hands.

00:39:10.660 | I heard people that I'm close with

00:39:11.980 | that are close to the organization

00:39:13.700 | and the company comment that they thought

00:39:15.680 | this is the most kind of scary, threatening,

00:39:18.420 | biggest threat to humanity

00:39:20.300 | is Google's control of DeepMind.

00:39:21.820 | And that was a naive kind of point of view,

00:39:24.660 | but it was one that was close,

00:39:25.900 | that was deeply held by a lot of people.

00:39:27.480 | So Reid Hoffman, Peter Thiel, Elon Musk,

00:39:29.820 | a lot of these guys funded the original

00:39:32.020 | kind of open AI business in 2015.

00:39:34.580 | And here's the link.

00:39:35.700 | So I'm putting it out here.

00:39:37.300 | You guys can pull up the original blog post.

00:39:39.140 | - Do all those people who donated get stock in?

00:39:42.300 | - So what happened was they were all in a non,

00:39:43.860 | it was all in a nonprofit.

00:39:45.040 | And then the nonprofit owned stock

00:39:46.700 | in a commercial business now.

00:39:48.280 | But your point is interesting

00:39:49.500 | because at the beginning,

00:39:50.940 | the idea was instead of having Google own all of this,

00:39:53.140 | we'll make it all available.

00:39:54.400 | And here's the statement from the original blog post in 2015,

00:39:58.260 | open AI is a nonprofit AI research company.

00:40:01.220 | Our goal is to advance digital intelligence

00:40:03.240 | in the way that is most likely to benefit humanity

00:40:05.180 | as a whole, unconstrained by a need

00:40:08.100 | to generate financial return.

00:40:10.180 | Since our research is free from financial obligations,

00:40:13.380 | we can better focus on a positive human impact.

00:40:16.060 | And they kind of went on and the whole thing

00:40:17.820 | about Sam, Greg, Elon, Reed, Jessica, Peter Thiel, AWS,

00:40:22.700 | YC are all donating to support open AI,

00:40:25.260 | including donations and commitments of over a billion

00:40:27.620 | dollars, although we expect that to only be a tiny fraction

00:40:32.060 | of what we will spend in the next few years,

00:40:35.140 | which is a really interesting kind of,

00:40:37.280 | if you look back historical perspective

00:40:39.100 | on how this thing all started seven years ago

00:40:41.380 | and how quickly it's evolved as you point out

00:40:44.060 | into the necessity to have a real commercial alignment

00:40:47.180 | to drive this thing forward without seeing any

00:40:49.220 | of these models open sourced.

00:40:50.700 | And during that same period of time,

00:40:53.420 | we've seen Google share AlphaFold and share a number

00:40:56.820 | of other predictive models and toolkits

00:40:59.740 | and make them publicly available

00:41:01.500 | and put them in Google's cloud.

00:41:04.220 | And so there's both kind of tooling and models

00:41:07.000 | and outputs of those models that Google has open sourced

00:41:09.860 | and made freely available.

00:41:11.220 | And meanwhile, open AI has kind of diverged

00:41:14.060 | into this deeply profitable, profit seeking

00:41:16.940 | kind of enterprise model.

00:41:18.340 | And when you invest in open AI,

00:41:20.940 | in the round that they did before,

00:41:22.980 | you could generate a financial return capped at 100X,

00:41:26.060 | which is still a pretty amazing financial return.

00:41:27.980 | You put a billion dollars in,

00:41:29.340 | you can make a hundred billion dollars.

00:41:30.660 | That's funding a real commercial endeavor at that point.

00:41:34.500 | - Well, and then to-

00:41:35.980 | - It is the most striking question about this whole thing,

00:41:38.260 | about what's going on in our AI.

00:41:40.220 | And it's one that Elon's talked about publicly

00:41:42.180 | and others have kind of sat on one side or the other,

00:41:45.060 | which is that AI offers a glimpse

00:41:49.300 | into one of the biggest

00:41:50.660 | and most kind of existential threats to humanity.

00:41:53.380 | And the question we're all gonna be tackling

00:41:56.180 | and the battle that's gonna be happening politically

00:41:58.580 | and regulatory wise,

00:42:00.060 | and perhaps even between nations in the years to come,

00:42:03.100 | is who owns the AI, who owns the models,

00:42:05.360 | what can they do with it?

00:42:06.340 | And what are we legally gonna be allowed to do with it?

00:42:08.460 | And this is a really important part of that story, yeah.

00:42:10.700 | - To build on what you're saying,

00:42:11.980 | I just put in PyTorch,

00:42:13.620 | people don't know that's another framework, P-Y-T-O-R-C-H.

00:42:18.260 | This was largely built inside of Facebook.

00:42:20.900 | And then Facebook said,

00:42:21.740 | "Hey, we want to democratize machine learning."

00:42:24.860 | And they made,

00:42:25.700 | and I think they put a bunch of executives,

00:42:27.500 | they may have even funded those executives

00:42:29.060 | to go work on this open source project.

00:42:31.940 | So they have a huge stake in this

00:42:33.700 | and they went very open source with it.

00:42:37.260 | And then TensorFlow,

00:42:38.460 | which you have an investment in Chamath,

00:42:42.020 | TensorFlow was inside of--

00:42:43.980 | - No, I don't have an investment in TensorFlow.

00:42:46.860 | We built--

00:42:47.700 | - No, TensorFlow, the public source came out of Google

00:42:49.300 | and then you invested in another company.

00:42:51.060 | - But we were building Silicon for machine learning.

00:42:53.100 | That's different.

00:42:54.260 | - Right.

00:42:55.100 | But it's based on TensorFlow.

00:42:56.900 | - No, no, no, no.

00:42:57.740 | The founder of this company was the founder of TensorFlow.

00:43:01.820 | - Oh, got it, okay.

00:43:02.860 | - Oh, sorry, not of TensorFlow, pardon me, of TPU,

00:43:06.420 | which was Google's internal Silicon

00:43:08.300 | that they built to accelerate TensorFlow.

00:43:11.260 | - Right.

00:43:12.100 | - If that makes sense.

00:43:13.460 | - And so that's the,

00:43:15.500 | I don't mean to be cynical about the whole project or not,

00:43:18.780 | it's just the confounding part of this

00:43:20.860 | and what is happening here.

00:43:21.900 | It reminds me, I don't know if you remember this,

00:43:24.300 | but Chamath and René allowed--

00:43:26.300 | - The biggest opportunity here is for Facebook.

00:43:28.020 | I mean, they need to get in this conversation, ASAP.

00:43:31.700 | I mean, to think that, like, look,

00:43:33.940 | PyTorch was like a pretty seminal piece of technology

00:43:37.180 | that a lot of folks in AI and machine learning

00:43:39.700 | were using for a long time.

00:43:41.100 | TensorFlow before that.

00:43:44.260 | And what's so funny about Google and Facebook

00:43:46.740 | is they're a little bit kind of like,

00:43:49.080 | they're not really making that much progress.

00:43:51.220 | I mean, Facebook released this kind of like

00:43:54.180 | rando version of AlphaFold recently.

00:43:57.180 | It's not that good.

00:43:58.480 | I think these companies really need to get these products

00:44:03.740 | in the wild as soon as possible.

00:44:05.560 | It cannot be the case that--

00:44:06.400 | - Is that, yeah.

00:44:07.460 | - You have to email people and get on some list,

00:44:09.620 | I mean, this is Google and Facebook, guys, come on.

00:44:12.100 | Get going. - This is the,

00:44:14.100 | I think the big innovation of OpenAI,

00:44:16.940 | SACS, to bring you in the conversation.

00:44:19.520 | They actually made an interface

00:44:21.300 | and let the public play with it

00:44:23.180 | to the tune of $3 million a day in cloud credits or costs.

00:44:27.380 | Which, you know-- - By the way, just on that,

00:44:30.340 | my son was telling me, he's like,

00:44:32.220 | "Hey, dad, do you want me to tell you

00:44:33.660 | when the best time to use ChatGPT is?"

00:44:35.460 | And I'm like, "Huh?"

00:44:36.340 | He's like, "Yeah, my friends and I,

00:44:38.500 | we've been using it so much,

00:44:39.460 | we know now when we can actually get resources."

00:44:43.420 | - Oh, wow. - And it's such

00:44:44.540 | an interesting thing where a 13-year-old kid knows

00:44:48.060 | when it's mostly compute-intensive that it's unusable

00:44:51.100 | and when to come back and use it.

00:44:52.380 | It's incredible. - When's the last time,

00:44:53.620 | SACS, that technology became this mainstream

00:44:56.380 | and captured people's imagination this broadly?

00:44:59.480 | Let's check GPT is. - It's been a while.

00:45:01.860 | I don't know, maybe the iPhone or something.

00:45:04.380 | Yeah, look, it's powerful.

00:45:06.060 | There's no question it's powerful.

00:45:07.100 | I mean, I'm of two minds about it

00:45:08.860 | because whenever something is the hype cycle,

00:45:11.420 | I just reflexively want to be skeptical of it.

00:45:14.000 | But on the other hand, we have made a few investments

00:45:16.660 | in this area.

00:45:18.100 | And I mean, I think it is powerful

00:45:20.420 | and it's going to be an enabler

00:45:21.980 | of some really cool things to come.

00:45:23.300 | There's no question about it.

00:45:24.340 | - I have two pieces of more insider information.

00:45:28.020 | One, I have a ChatGPT iOS app on my phone.

00:45:31.420 | One of the nice folks at OpenAI

00:45:33.580 | included me in the test flight.

00:45:35.540 | And it's the simplest interface you've ever seen,

00:45:37.740 | but basically you type in your question,

00:45:39.700 | but it keeps your history.

00:45:41.180 | And then you can search your history.

00:45:42.300 | So it looks, SACS, like you're in iMessage, basically.

00:45:46.180 | And it has your threads.

00:45:47.580 | And so I asked him,

00:45:48.420 | "Hey, what are the best restaurants in Yawnville?"

00:45:51.100 | A town near Napa.

00:45:53.380 | And then I said, "Which one has the best duck?"

00:45:55.540 | And it literally like gave me a great answer.

00:45:57.780 | And then I thought, "Wait a second,

00:45:58.620 | "why is this not using a Siri or Alexa-like interface?

00:46:02.840 | "And then why isn't it, oh, here's a video of it."

00:46:04.940 | And I gave the video to Nick.

00:46:06.260 | - By the way, Jason, this, what you're doing right now

00:46:08.580 | is you're creating a human feedback

00:46:10.860 | reinforcement learning pipeline for ChatGPT.

00:46:13.780 | So just the fact that you asked that question,

00:46:16.620 | and over time, if ChatGPT has access to your GPS information

00:46:21.260 | and then knows that you went to restaurant A versus B,

00:46:24.460 | it can intuit.

00:46:25.420 | And it may actually prompt you to ask,

00:46:27.880 | "Hey, Jason, we noticed you were in the area.

00:46:30.360 | "Did you go to Bottega?

00:46:31.420 | "If you did, how would you rate it one through five?"

00:46:34.020 | That reinforcement learning now allows the next person

00:46:36.780 | that asks, "What are the top five restaurants?"

00:46:38.540 | to say, "Well, over a thousand people

00:46:40.880 | "that have asked this question,

00:46:41.800 | "here's actually the best answer,"

00:46:43.860 | versus a generic rank of the open web,

00:46:46.460 | which is what the first data set is.

00:46:48.380 | That's what's so interesting about this.

00:46:50.300 | So this is why, if you're a company

00:46:52.460 | that already owns the eyeballs,

00:46:54.460 | you have to be running to get this stuff out there.

00:46:57.740 | - Well, and then this answer, you cited Yelp.

00:47:01.100 | Well, this is the first time I've actually seen

00:47:03.140 | a Chatshippeetee site,

00:47:04.740 | and this is, I think, a major legal breakthrough.

00:47:06.780 | It didn't put a link in,

00:47:08.500 | but if it's gonna use Yelp's data,

00:47:10.780 | I don't know if they have permission from Yelp,

00:47:12.180 | but it's quoting Yelp here,

00:47:13.620 | it should link to French Lounge, Bottega, and Bouchon.

00:47:16.300 | Bouchon actually has the best duck confit, for the record.

00:47:19.620 | And I did have that duck,

00:47:20.840 | so I asked this afterwards to see,

00:47:23.060 | you know, in a scenario like this.

00:47:24.720 | But it could also, if I was talking to it,

00:47:27.060 | I could say, "Hey, which one has availability

00:47:28.900 | "this afternoon or tomorrow for dinner?"

00:47:31.980 | And make the phone call for me,

00:47:33.180 | like Google Assistant does,

00:47:34.380 | or any number of next tasks. - I was thinking about--

00:47:37.980 | - This was an incredibly powerful display

00:47:40.740 | in a 1.0 product.

00:47:42.060 | - I was thinking about what you said last week,

00:47:43.860 | and I thought back to the music industry

00:47:47.060 | in the world of Napster.

00:47:50.660 | And what happened was, there was a lot of musicians,

00:47:54.280 | I think Metallica being the most famous one,

00:47:57.480 | famously suing Napster, because it was like,

00:48:00.580 | "Hey, listen, you're allowing people to take my content,

00:48:04.080 | "which they would otherwise pay for.

00:48:05.480 | "There's economic damage that I can measure."

00:48:08.180 | That legal argument was meaningful enough

00:48:10.200 | that ultimately Napster was shut down.

00:48:11.880 | Now, there were other versions of that,

00:48:14.140 | that folks created, including us at Winamp.

00:48:16.260 | We created a headless version of that.

00:48:18.700 | But if you translate that problem set here,

00:48:21.820 | is there a claim that Yelp can make in this example,

00:48:25.700 | that they're losing money?

00:48:27.900 | That if you were going through Google,

00:48:30.520 | or if you were going through their app,

00:48:33.740 | there's the sponsored link revenue

00:48:35.560 | and the advertising revenue that they would have got

00:48:37.660 | that they wouldn't get from here.

00:48:39.060 | Now, that doesn't mean that ChatGPT can't figure that out,

00:48:42.420 | but it's those kinds of problems

00:48:43.740 | that are gonna be a little thorny in these next few years

00:48:47.100 | that have to really get figured out.

00:48:49.560 | This is a very--

00:48:51.420 | - If you were a human reading every review on Yelp

00:48:54.480 | about duck, then you could write a blog post

00:48:58.400 | in which you say, "Many reviewers on Yelp

00:49:00.520 | "say that Bouchon is the best duck."

00:49:02.440 | So the question is, is GPT held to that standard?

00:49:06.760 | - Yeah, exactly. - Or is it,

00:49:07.700 | or something different?

00:49:09.640 | And is linking to it enough?

00:49:12.200 | - This is the question that I'm asking.

00:49:13.480 | I don't know.

00:49:14.460 | It should be, I'll argue it should be,

00:49:16.680 | because if you look at the four-part test for fair use,

00:49:19.480 | which I had to go through

00:49:20.320 | because blogging had the same issue.

00:49:21.600 | We would write a blog post

00:49:23.720 | and we would mention Walt Mossberg's review of a product

00:49:26.080 | and somebody else's.

00:49:26.920 | And then people would say,

00:49:27.740 | "Oh, I don't need to read Walt Mossberg's

00:49:29.120 | "Indiana Wall Street Journal subscription."

00:49:30.720 | And we'd say, "Well, we're doing an original work.

00:49:32.160 | "We're comparing two or three different."

00:49:34.200 | You know, human is comparing two or three different reviews

00:49:39.160 | and we're adding something to it.

00:49:40.600 | It's not interfering with Walt Mossberg's ability

00:49:45.200 | to get subscribers in the Wall Street Journal.

00:49:47.440 | But the effect on the potential market

00:49:49.680 | is one of the four tests.

00:49:50.960 | And just reading from Stanford's quote on fair use,

00:49:54.320 | "Another important fair use factor

00:49:55.560 | "is whether your use deprives the copyright owner of income

00:49:59.100 | "or undermines a new or potential market

00:50:01.340 | "for the copyrighted work.

00:50:02.980 | "Depriving a copyright owner of income

00:50:05.200 | "is very likely to trigger a lawsuit."

00:50:07.320 | This is true even if you are not competing directly

00:50:09.560 | with the original work.

00:50:11.400 | And we'll put the link to Stanford here.

00:50:13.240 | This is the key issue.

00:50:14.380 | And I would not use Yelp.

00:50:16.160 | In this example, I would not open the Yelp app.

00:50:18.880 | Yelp would get no commerce and Yelp would lose this.

00:50:20.960 | So, ChatGPT and all these services must use citations

00:50:25.960 | of where they got the original work.

00:50:27.500 | They must link to them and they must get permission.

00:50:29.380 | That's where this is all gonna shake out.

00:50:31.080 | And I believe that's--

00:50:31.920 | - Using permission?

00:50:32.740 | Well, forget about permission.

00:50:33.580 | I mean, you can't get a big enough data set

00:50:35.560 | if you have to get permission in advance, right?

00:50:38.160 | You have to go out and negotiate.

00:50:39.880 | - It's gonna be the large data sets.

00:50:41.400 | Quora, Yelp, the App Store reviews, Amazon's reviews.

00:50:46.520 | So, there are large corpuses of data that you would need.

00:50:49.760 | Like Craigslist has famously never allowed anybody

00:50:52.080 | to scrape Craigslist.

00:50:53.280 | The amount of data inside Craigslist,

00:50:55.480 | but one example of a data set,

00:50:57.280 | would be extraordinary to build ChatGPT on.

00:50:59.920 | ChatGPT is not allowed to because,

00:51:02.000 | as you brought up robots.txt last week,

00:51:04.340 | there's gonna need to be an AI.txt.

00:51:08.240 | Are you allowed to use my data set in AI?

00:51:10.240 | And how will I be compensated for it?

00:51:13.700 | I'll allow you to use Craigslist,

00:51:15.800 | but you have to link to the original post,

00:51:18.560 | and you have to note that.

00:51:20.480 | - The other gray area that isn't there today,

00:51:22.640 | but may emerge, is when section 230 gets rewritten.

00:51:27.560 | Because if they take the protections away

00:51:29.560 | for the Facebook and the Googles of the world,

00:51:31.260 | for the basically for being an algorithmic publisher,

00:51:34.560 | and saying an algorithm is equivalent to a publisher.

00:51:37.580 | What it essentially saying is that an algorithm

00:51:40.560 | is kind of like doing the work of a human

00:51:42.700 | in a certain context.

00:51:43.600 | And I wonder whether that's also an angle here,

00:51:45.640 | which now this algorithm, which today,

00:51:48.280 | David, you said the example,

00:51:49.920 | I read all these blog posts, I write something.

00:51:52.520 | But if an algorithm does it,

00:51:53.880 | maybe can you then say, no, actually there was intent there

00:51:56.880 | that's different than if a human were to do it?

00:51:58.840 | I don't know.

00:51:59.660 | My point is, very complicated issues

00:52:02.800 | that are gonna get sorted out.

00:52:04.680 | And I think the problem with the hype cycle

00:52:08.200 | is that you're gonna have to marry it

00:52:09.500 | with an economic model for VCs to really make money.

00:52:12.980 | And right now there's just too much betting on the come.

00:52:15.020 | So to the extent you're going to invest,

00:52:17.260 | it makes sense that you put money into open AI

00:52:19.400 | because that's safe.

00:52:20.440 | Because the economic model of how you make money

00:52:23.720 | for everybody else is so unclear.

00:52:25.840 | - No, it's clear actually, I have it for business.

00:52:28.880 | I just signed up for chat GPT premium.

00:52:31.280 | They had a survey that they shared on their discord server.

00:52:34.800 | And I filled out the survey

00:52:35.880 | and they did a price discovery survey, Freeberg.

00:52:38.960 | What's the least you would pay, the most you would pay,

00:52:41.320 | what would be too cheap of a price for chat GPT pro,

00:52:44.520 | and what would be too high of a price?

00:52:45.960 | I put in like 50 bucks a month would be what I would pay.

00:52:49.080 | But I was just thinking, imagine chat GPT

00:52:51.080 | allowed you, Freeberg, to have a Slack channel

00:52:53.800 | called research.

00:52:55.040 | And you could go in there or anytime you're in Slack,

00:52:57.620 | you do slash chat or slash chat GPT.

00:53:00.600 | And you say slash chat GPT, tell me,

00:53:03.200 | what are the venues available in,

00:53:05.520 | which we did this actually for,

00:53:07.520 | I did this for venues for all in some way.

00:53:10.280 | I can say, what are the venues that seat

00:53:11.400 | over 3000 people in Vegas?

00:53:14.440 | And it just gave us the answer.

00:53:15.760 | Okay, well, that was the job of the local event planner.

00:53:20.760 | They had that list.

00:53:21.840 | Now you can pull that list

00:53:23.280 | from a bunch of different sources.

00:53:24.720 | I mean, what would you pay for that?

00:53:26.880 | A lot.

00:53:27.720 | - Well, I think one of the big things that's happening

00:53:31.120 | is all the old business models don't make sense anymore.

00:53:36.120 | In a world where the software is no longer just doing

00:53:40.040 | what it's done for the last 60 years,

00:53:42.440 | which is what is historically defined

00:53:44.480 | as information retrieval.

00:53:46.320 | So you have this kind of hierarchical storage of data

00:53:50.320 | that you have some index against,

00:53:52.840 | and then you go and you search and you pull data out,

00:53:55.400 | and then you present that data back to the customer

00:53:57.600 | or the user of the software.

00:53:59.200 | And that's effectively been how all kind of data

00:54:04.200 | has been utilized in all systems

00:54:06.700 | for the past 60 years in computing.

00:54:09.600 | Largely, what we've really done

00:54:11.600 | is kind of built an evolution of application layers

00:54:14.080 | or software tools to interface with the fetching of that data,

00:54:17.800 | the retrieval of that data, and the display of that data.

00:54:20.400 | But what these systems are now doing,

00:54:22.600 | what AI type systems or machine learning systems now do,

00:54:25.880 | is the synthesis of that data

00:54:27.900 | and the representation of some synthesis of that data

00:54:31.720 | to you, the user, in a way that doesn't necessarily

00:54:34.760 | look anything like the original data

00:54:36.860 | that was used to make that synthesis.

00:54:39.420 | And that's where business models like a Yelp, for example,

00:54:42.380 | or like a web crawler that crawls the web

00:54:44.740 | and then presents webpage directories to you.

00:54:47.600 | Those sorts of models no longer make sense

00:54:49.740 | in a world where the software,

00:54:50.820 | the signal to noise is now greater.

00:54:52.840 | The signal is greater than the noise

00:54:54.620 | in being able to present to you a synthesis of that data

00:54:57.780 | and basically resolve what your objective is

00:55:01.200 | with your own consumption and interpretation of that data,

00:55:04.240 | which is how you've historically used these systems.

00:55:06.740 | And I think that's where there's,

00:55:08.260 | going back to the question of the hype cycle,

00:55:10.660 | I don't think it's about being a hype cycle.

00:55:12.500 | I think it's about the investment opportunity

00:55:15.220 | against fundamentally rewriting all compute tools.

00:55:18.420 | Because if all compute tools ultimately can use

00:55:21.280 | this capability in their interface and in their modeling,

00:55:24.380 | then it very much changes everything.

00:55:26.320 | And one of the advantages that I think

00:55:28.300 | businesses are going to latch onto,

00:55:29.980 | which we talked about historically,

00:55:31.660 | is novelty in their data,

00:55:33.340 | in being able to build new systems and new models

00:55:36.700 | that aren't generally available.

00:55:38.980 | - Example?

00:55:40.140 | - In biotech and pharma, for example,

00:55:42.860 | having screening results from very expensive experiments

00:55:47.860 | and running lots of experiments

00:55:49.620 | and having a lot of data against those experiments

00:55:51.820 | gives a company an advantage

00:55:55.140 | in being able to do things like drug discovering.

00:55:57.140 | We're going to talk about that in a minute,

00:55:59.140 | versus everyone using publicly known screening libraries

00:56:02.180 | or publicly available protein modeling libraries,

00:56:05.580 | and then screening against those.

00:56:06.900 | And then everyone's got the same candidates

00:56:08.260 | and the same targets and the same clinical objectives

00:56:11.260 | that they're going to try and resolve from that output.

00:56:13.660 | So I think novelty in data

00:56:16.300 | is one way that advantage arises.

00:56:19.980 | But really, that's just kind of, where's there an edge?

00:56:23.580 | But fundamentally, every business model can

00:56:26.100 | and will need to be rewritten

00:56:27.900 | that's dependent on the historical,

00:56:31.460 | on the legacy of information retrieval

00:56:34.220 | as the core of what computing is used to do.

00:56:37.260 | - Sax, on my other podcast,

00:56:38.460 | I was having a discussion with Molly

00:56:39.780 | about the legal profession.

00:56:41.640 | What impact would it be if ChantGPT took every court case,

00:56:46.500 | every argument, every document,

00:56:48.820 | and somebody took all of those legal cases

00:56:53.180 | on the legal profession,

00:56:55.220 | and then the filing of a lawsuit,

00:56:56.800 | the defending of a lawsuit,

00:56:59.060 | public defenders, prosecutors,

00:57:01.820 | what data could you figure out?

00:57:03.780 | Like, just to think of the recent history,

00:57:06.420 | you look at Chesa Boudin,

00:57:07.900 | you could literally take every case,

00:57:10.140 | every argument he did, put it through it,

00:57:12.460 | and say, you know, versus an outcome in another state,

00:57:14.980 | and you could figure out what's actually going on

00:57:18.940 | with this technology.

00:57:19.780 | What impact did this have on the legal field?

00:57:22.020 | You are a non-practicing attorney,

00:57:24.020 | you have a legal degree.

00:57:25.380 | - I never practiced,

00:57:27.180 | other than one summer at a law firm.

00:57:29.060 | But no, I think--

00:57:29.900 | - Did you pass the bar?

00:57:30.740 | - What's that?

00:57:31.560 | I did pass the bar, yes.

00:57:32.500 | - First try.

00:57:33.900 | - Yes, of course.

00:57:34.740 | - First try, of course, yeah, come on.

00:57:35.980 | - You gotta be kind of dumb to fail the bar exam.

00:57:36.820 | - Look, it's like a rolling in science,

00:57:38.340 | yes, I went to Stanford, dude.

00:57:40.300 | I'm rated 1800.

00:57:41.140 | - It took two days of studying, come on.

00:57:42.340 | - I may not have passed the bar,

00:57:43.580 | but I know a little shit,

00:57:44.900 | enough to know that you can't legally--

00:57:46.620 | - I would be curious in terms of

00:57:49.620 | a very common question

00:57:52.140 | that an associate at a law firm would get asked

00:57:55.100 | would be something like,

00:57:56.580 | you know, summarize the legal precedence

00:57:58.660 | in favor of X, right?

00:58:01.860 | And you could imagine GPT doing that like instantly.

00:58:06.380 | Now, I think that the question about that,

00:58:08.380 | I think there's two questions.

00:58:09.360 | One is, can you prompt GPT in the right way

00:58:12.340 | to get the answer you want?

00:58:13.780 | And I think, you know, Chamath,

00:58:14.860 | you shared a really interesting video

00:58:16.780 | showing that people are developing some skills

00:58:19.140 | around knowing how to ask GPT questions in the right way.

00:58:23.740 | Is that prompt engineering?

00:58:25.460 | Why?

00:58:26.300 | 'Cause GPT's a command line interface.

00:58:27.700 | So if you ask GPT a simple question about

00:58:30.300 | what's the best restaurant in, you know, Napa,

00:58:33.020 | it knows how to answer that.

00:58:34.200 | But there are much more complicated questions

00:58:36.560 | that you kind of need to know

00:58:37.900 | how to prompt it in the right way.

00:58:39.060 | So it's not clear to me that a command line interface

00:58:41.740 | is the best way of doing that.

00:58:42.740 | I could imagine apps developing

00:58:44.700 | that create more of like a GUI.

00:58:46.320 | So we're an investor, for example, in Copy AI,

00:58:48.320 | which is doing this for copywriters and marketers,

00:58:51.060 | helping them write blog posts and emails.

00:58:53.780 | And so, you know, imagine putting that like, you know,

00:58:57.820 | GUI on top of Chad GPT.

00:59:00.020 | They've already been kind of doing this.

00:59:01.940 | So I think that's part of it.

00:59:03.980 | I think the other part of it is on the answer side,

00:59:07.500 | you know, how accurate is it?

00:59:09.780 | Because in some professions,

00:59:12.140 | having 90 or 95 or 99% accuracy is okay.

00:59:16.640 | But in other professions, you need six nines accuracy,

00:59:19.620 | meaning 99.9999% accuracy.

00:59:23.780 | Okay, so I think for a lawyer going into court,

00:59:28.340 | you know, you probably need, I don't know,

00:59:31.180 | it depends on the question.

00:59:32.020 | - Yeah, it's a parking ticket versus a murder trial

00:59:34.900 | is two very different things.

00:59:36.060 | - Yeah, exactly.

00:59:36.900 | So is 99% accuracy good enough?

00:59:39.420 | Is 95% accuracy good enough?

00:59:41.780 | I would say probably for a court case,

00:59:44.740 | 95% is probably not good enough.

00:59:46.120 | I'm not sure GPT is at even 95% yet.

00:59:49.060 | But could it be helpful?

00:59:51.300 | Like could the associates start with Chad GPT,

00:59:55.300 | get an answer and then validate it?

00:59:57.860 | Probably, yeah.

00:59:58.860 | - If you had a bunch of associates

01:00:00.900 | bang on some law model for a year,

01:00:04.260 | again, that's that reinforcement learning

01:00:05.940 | we just talked about.

01:00:07.300 | I think you'd get precision recall off the charts

01:00:09.740 | and it would be perfect.

01:00:11.420 | By the way, just a cute thing.

01:00:13.060 | I don't know if you guys got this email.

01:00:14.940 | It came about an hour ago from Reid Hoffman.

01:00:17.180 | And Reid said to me, "Hey Chamath,

01:00:18.420 | I created Fireside Chatbots,

01:00:22.540 | a special podcast mini series

01:00:25.140 | where I will be having a set of conversations

01:00:27.460 | with Chad GPT."

01:00:28.820 | So you can go to YouTube by the way

01:00:30.420 | and see Reid having, and he's a very smart guy,

01:00:33.420 | so this should be kind of cool.

01:00:35.740 | And by the way, Chad GPT will have an AI generated voice

01:00:39.260 | powered by the text to speech platform, play.ht.

01:00:44.140 | Go to YouTube if you want to see Reid

01:00:46.420 | have a conversation with Chad GPT.

01:00:50.660 | - Well, I mean, Chamath, we have a conversation

01:00:52.540 | with the two Davids every week.

01:00:54.100 | What's the difference?

01:00:54.940 | We know how this is gonna turn out.

01:00:57.580 | - Hey, but actually, so synthesizing Chamath's point

01:01:00.660 | about reinforcement learning

01:01:01.980 | with something you said, Jay Cal, in our chat,

01:01:04.420 | which I actually thought was pretty smart.

01:01:06.740 | - Well, that's a first.

01:01:07.860 | - Yeah, so I'm gonna give you credit here

01:01:09.700 | 'cause I don't think you've said it on this episode,

01:01:12.660 | which is you said that these open AI capabilities

01:01:16.340 | are eventually gonna become commoditized

01:01:18.380 | or certainly much more widely available.

01:01:21.020 | I don't know if that means

01:01:21.980 | that they'll be totally commoditized

01:01:23.220 | or there'll be four players,

01:01:24.420 | but there'll be multiple players that offer them.

01:01:26.660 | And you said the real advantage will come from

01:01:29.940 | applications that are able to get ahold

01:01:31.980 | of proprietary data sets

01:01:34.140 | and then use those proprietary data sets

01:01:36.860 | to generate insights.

01:01:37.940 | And then layering on what Chamath said

01:01:39.180 | about reinforcement learning,

01:01:40.700 | if you can be the first out there

01:01:42.700 | in a given vertical with a proprietary data set,

01:01:45.380 | and then you get the advantage,

01:01:47.020 | the moat of reinforcement learning,

01:01:49.340 | that would be the way to create,

01:01:50.540 | I think, a sustainable business.

01:01:51.980 | - Just to build on what you said,

01:01:53.540 | this week is the J.P. Morgan Conference.

01:01:56.820 | Freeberg mentioned it last week.

01:01:58.620 | I had dinner on Wednesday

01:01:59.660 | with this really interesting company based in Zurich.

01:02:03.180 | And what they have is basically a library of ligands, right?

01:02:07.700 | And so these ligands are used as a substrate

01:02:10.540 | to deliver all kinds of molecules inside the body.

01:02:13.220 | And what's interesting is that they have a portfolio

01:02:15.900 | of like a thousand of these,

01:02:17.620 | but really what they have

01:02:19.180 | is they have all the nuclear medicine

01:02:21.540 | about whether it works.

01:02:22.980 | So, you know, they target glioblastoma, glioblastoma.

01:02:26.740 | And so all of a sudden they can say,

01:02:28.500 | "Well, this ligand can actually cross the blood-brain barrier

01:02:31.020 | "and get to the brain."

01:02:32.460 | They have an entire data set of that,

01:02:34.100 | and a whole bunch of nuclear imagery around that.

01:02:36.180 | They have something for soft-cell carcinoma.

01:02:38.500 | So then they have that data set.

01:02:40.260 | So to your point, that's really valuable

01:02:42.740 | because that's real work that Google or Microsoft

01:02:47.260 | or OpenAI won't do, right?

01:02:51.100 | And if you have that and you bring it to the problem,

01:02:53.940 | you can probably make money.

01:02:55.540 | You know, there's a business there to be built.

01:02:57.780 | - Just building on this conversation,

01:02:59.740 | I just realized like a great prompt engineer

01:03:02.900 | is gonna become a title and an actual skill,

01:03:05.100 | the ability to interface with these AIs.

01:03:06.980 | - Here you go, coding school 2.0.

01:03:08.820 | - And that prompt engineer, well, no, a prompt engineer,

01:03:11.780 | somebody who is very good at talking to these, you know,

01:03:15.780 | instances and maximizing the result for them

01:03:18.740 | and refining the results for them,

01:03:20.340 | just like a detective who asks great questions,

01:03:23.100 | that person is gonna be 10 or 20 times more valuable.

01:03:27.300 | They could be the proverbial 10X engineer

01:03:30.180 | in the future of as in a company.

01:03:33.340 | And as we talk about austerity and doing more with less

01:03:36.220 | and the 80% less people running Twitter now

01:03:39.140 | or Amazon laying off 18,000 people,

01:03:41.740 | Salesforce laying off 8,000,

01:03:43.180 | Facebook laying off 10 and probably another 10,000,

01:03:46.220 | what catalytic effect like could this have?

01:03:50.180 | We could be sitting here in three or four or five years

01:03:51.980 | and instead of running a company like Twitter

01:03:54.140 | with 80% less people,

01:03:55.180 | maybe you could run it with 98% less people.

01:03:57.580 | - Look, I think directionally, it's the right statement.

01:04:00.660 | I mean, you know, I've made the statement a number of times

01:04:03.780 | that I think we move from this idea of creator economy

01:04:07.540 | to narrator economy,

01:04:08.700 | where historically it was kind of labor economy,

01:04:11.060 | where humans use their physical labor to do things.

01:04:13.780 | Then we were knowledge workers,

01:04:15.100 | we used our brains to make things.

01:04:19.380 | And then ultimately we kind of, I think,

01:04:21.820 | resolve to this narrator economy

01:04:23.580 | where the way that you kind of can state intention

01:04:26.260 | and better manipulate the tools

01:04:28.380 | to drive your intentional outcome,

01:04:30.180 | the more successful you're gonna be.

01:04:31.140 | And you can kind of think about this

01:04:32.220 | as being the artist of the past.

01:04:34.140 | Da Vinci was, what made him so good

01:04:37.940 | was he was technically incredible

01:04:40.260 | at trying to reproduce a photographic like imagery

01:04:44.060 | using paint.

01:04:46.380 | And there's these really great kind of museum exhibits

01:04:48.660 | on how he did it using these really interesting

01:04:50.980 | kind of like split mirror systems.

01:04:55.020 | And then the better, the artist of the 21st century,

01:04:58.020 | the 20th century was the best user of Adobe Photoshop.

01:05:01.020 | And that person is not necessarily the best painter.

01:05:03.780 | And the artist of the 22nd century

01:05:06.540 | isn't gonna look like the Photoshop expert.

01:05:09.660 | And it's not gonna look like the painter,

01:05:12.020 | it's gonna look like something entirely different.

01:05:13.980 | It could be who's got the most creative imagination

01:05:16.900 | in driving the software to drive new outcomes.

01:05:19.140 | And I think that the same analogy can be used

01:05:21.020 | across every market and every industry.

01:05:23.180 | However, one thing to note, Jekyll,

01:05:25.220 | it's not about austerity

01:05:27.260 | 'cause the Luddite argument is when you have new tools

01:05:31.220 | and you get more leverage from those tools,

01:05:33.380 | you have less work for people to do

01:05:35.060 | and therefore everyone suffers.

01:05:36.780 | The reality is new work emerges

01:05:38.820 | and new opportunities emerge.

01:05:40.460 | And we level up as a species.

01:05:42.380 | And when we level up, we all kind of fill the gaps

01:05:44.460 | and expand our productivity and our capability set.

01:05:47.380 | - I thought what Jekyll was saying was more

01:05:49.020 | that Google will be smaller

01:05:50.500 | didn't mean that the pie wouldn't grow.

01:05:52.100 | It's just that that individual company is run differently,

01:05:55.540 | but there will be hundreds of more companies

01:05:57.780 | or thousands more, millions more.

01:05:59.220 | - Yeah, that's sort of,

01:06:00.340 | I have an actual punch up for you.

01:06:02.900 | Instead of narrative, it's the conductor economy.

01:06:05.340 | It's you're conducting a symphony.

01:06:07.980 | - Ooh, a punch up.

01:06:09.620 | - Punch up there.

01:06:10.460 | But I do think like we're gonna,

01:06:11.740 | there's gonna be somebody who's sitting there like,

01:06:14.020 | remember Tom Cruise in Minority Report

01:06:15.660 | as a detective was moving stuff around

01:06:17.460 | with a interface in, you know,

01:06:20.140 | with the gloves and everything.

01:06:21.900 | This is kind of that manifested.

01:06:23.860 | You could, even if you're not an attorney,

01:06:25.380 | you can say, hey, I wanna sue this company

01:06:27.380 | for copyright infringement, give me my best arguments.

01:06:29.940 | And then on the other side say,

01:06:30.980 | hey, I wanna know what the next three features

01:06:32.860 | I should put into my product is.

01:06:34.660 | Can you examine who are my top 20 competitors?

01:06:37.180 | And then who have they hired in the last six months?

01:06:39.100 | And what are those people talking about on Twitter?

01:06:40.940 | You can have this conductor, you know,

01:06:43.660 | who becomes really good at that.

01:06:45.220 | - Well, the leveling up--

01:06:46.060 | - Massive amounts of tasks.

01:06:48.260 | - Yeah, the leveling up that happens

01:06:49.660 | in the book Ender's Game, I think is a good example

01:06:52.620 | of this where the guy goes through the entire

01:06:54.860 | kind of ground up and then ultimately

01:06:56.340 | he's commanding armies of spaceships in space

01:06:59.580 | and his orchestration of all of these armies

01:07:02.380 | is actually the skillset that wins the war.

01:07:04.860 | Yeah.

01:07:05.700 | - You predicted that there would be like all these people

01:07:07.700 | that create these next gen forms of content.

01:07:10.700 | But I think this Reid Hoffman thing could be pretty cool.

01:07:13.140 | Like what if he wins a Grammy for his, you know,

01:07:16.260 | computer created podcast miniseries?

01:07:18.860 | That's really cool.

01:07:19.700 | - The thing I'm really excited about,

01:07:21.260 | when's the first AI novel gonna get published

01:07:23.380 | by a major publisher?

01:07:24.380 | I think it happens this year.

01:07:25.740 | When's the first AI symphony gonna get performed

01:07:27.820 | by a major symphony orchestra?

01:07:29.420 | And when's the first AI generated screenplay

01:07:31.540 | get turned into an AI generated 3D movie that we all watch?

01:07:34.900 | And then the more exciting one I think is

01:07:36.340 | when do we all get to make our own AI video game

01:07:38.980 | where we instruct the video game platform

01:07:40.900 | what world we wanna live in?

01:07:42.140 | I don't think that's happening

01:07:43.060 | for the next three or four years,

01:07:44.460 | but when it does, I think everyone's got

01:07:46.020 | these new immersive environments that they can live in.

01:07:48.180 | - I have a question.

01:07:49.180 | When, you know, when you think--

01:07:50.020 | - When I say live in, I mean video game wise.

01:07:51.620 | Yeah, sorry, go ahead.

01:07:52.460 | - When you have these computer systems,

01:07:54.860 | just like to use a question of game theory for a second,

01:07:57.380 | they're, these models are iterating rapidly.

01:07:59.980 | These are all mathematical models.

01:08:01.180 | So inherent in, let's just say this,

01:08:04.700 | the perfect answer, right?

01:08:06.260 | Like if you had perfect precision recall,

01:08:10.540 | if multiple models get there at a system-wide level,

01:08:13.340 | everybody is sort of like,

01:08:15.020 | they get to the game theory optimal.

01:08:17.300 | They're all at Nash equilibrium, right?

01:08:19.060 | All these systems working at the same time.

01:08:21.580 | Then the real question would then be

01:08:24.260 | what the hell do you do then?

01:08:26.180 | 'Cause if you keep getting the same answer,

01:08:27.820 | if everybody then knows how to ask the exact right question

01:08:30.980 | and you start to go through these iterations

01:08:32.660 | where you're like, maybe there is a dystopian hellscape

01:08:35.340 | where there are no jobs.

01:08:37.020 | Maybe that's the Elon world, which is you can,

01:08:40.060 | you can recursively find a logical argument

01:08:44.340 | where there is no job that's possible, right?

01:08:47.940 | And now I'm not saying that that path is the likely path,

01:08:50.940 | but I'm saying it is important to keep in mind

01:08:52.620 | that that path of outcomes is still very important

01:08:55.860 | to keep in the back of our mind

01:08:57.300 | as we figure these things out.

01:08:58.660 | - Well, Freeberg, you were asking before about this,

01:09:01.180 | like, will more work be created?

01:09:03.700 | Of course, artistic pursuits,

01:09:05.420 | podcasting is a job now, being an influencer is a job,

01:09:08.540 | yada, yada, new things emerge in the world.

01:09:10.420 | But here in the United States in 1970,

01:09:13.460 | I'm looking at Fred, I'm looking at the St. Louis Fed,

01:09:16.820 | 1970, 26.4% of the country was working in a factory,

01:09:21.820 | was working in manufacturing.

01:09:23.820 | You want to guess what that is in 2012?

01:09:27.800 | - Sorry, what percentage?

01:09:29.240 | - It was 26% in 1970.

01:09:32.140 | And in 2015, when they stopped the percentage

01:09:34.900 | in manufacturing, I say it's, they discontinued this,

01:09:37.020 | it was a 10.

01:09:38.500 | So it's possible we could just see, you know,

01:09:42.020 | the concept of office work,

01:09:44.060 | the concept of knowledge work is going to follow,

01:09:47.260 | pretty inevitable, the path of manufacturing.

01:09:50.580 | That seems like a pretty logical theory or no?

01:09:52.880 | - I think we should move on, but yes.

01:09:56.820 | - Okay, so how would we like to ruin the show now?

01:09:59.500 | Should we talk about Biden and the documents

01:10:02.460 | and ruin the show with political talk?

01:10:04.780 | Or should we talk about?

01:10:07.480 | Since it's been such a great episode so far,

01:10:10.280 | what do we want to talk about next?

01:10:12.160 | I'll give you a couple of choices.

01:10:13.000 | - I know what you don't want to talk about.

01:10:14.880 | (laughing)

01:10:16.560 | - Some of these network guys are talking about productivity.

01:10:19.200 | - Give it to him, give it to him.

01:10:20.880 | - Wait, we all know, we all know J. Cal.

01:10:23.920 | - Apple cash, time-max, hourglasses?

01:10:24.760 | - Hold on a second, we all know J. Cal,

01:10:26.600 | that according to you, when a president

01:10:29.200 | is in possession of classified documents in his home,

01:10:33.760 | that apparently have been taken in an unauthorized manner,

01:10:36.760 | basically stolen, he should have his home raided by the FBI.

01:10:40.040 | - Almost, close, close.

01:10:42.640 | So anyway, Biden, as of the taping of this,

01:10:49.400 | has now said there's a third batch of classified documents.

01:10:54.160 | This group, I guess there was one at an office,

01:10:56.760 | one at a library, now this third group is in his garage

01:10:59.200 | with his Corvette, certainly not looking good.

01:11:01.960 | - Well, they say in his defense,

01:11:04.240 | they say the garage was locked,

01:11:05.640 | meaning that you could use a garage door opener

01:11:08.400 | to open and close it.

01:11:11.000 | It was locked when it went closed.

01:11:12.320 | - So pretty much as secure as the documents at Mar-a-Lago,

01:11:15.840 | same equivalency.

01:11:16.880 | - No, no, no, actually, I mean, just to be perfectly fair,

01:11:19.880 | the documents at Mar-a-Lago were locked in a basement.

01:11:22.200 | The FBI came, checked it out,

01:11:24.160 | said we'd like you to lock those up, they locked 'em up.

01:11:26.200 | So a little safer than being in a garage with a Corvette.

01:11:31.280 | - But functionally the same, functionally the same.

01:11:34.480 | - The only difference here would be what,

01:11:36.680 | Sachs, when you look at these two cases?

01:11:39.320 | - Well, that in one case, Merrick Garland

01:11:41.800 | is appointed an independent counsel to investigate Trump,

01:11:44.640 | and there's no such a special counsel

01:11:47.120 | or investigator appointed to investigate Biden.

01:11:49.640 | I mean, these things are functionally the same.

01:11:50.480 | - Didn't he put somebody on it, though?

01:11:52.280 | Wait, did he put somebody on it?

01:11:54.120 | - No, I don't think they've appointed

01:11:55.280 | a special counsel yet.

01:11:56.480 | - No, they did.

01:11:57.320 | As of an hour ago, a special counsel was appointed.

01:12:00.320 | - Okay.

01:12:01.160 | - Did that just happen?

01:12:02.000 | - Yeah, one hour ago.

01:12:03.000 | Robert Herr is his name.

01:12:03.960 | - Okay, I guess there are real questions to look into here.

01:12:07.080 | The documents apparently were moved twice.

01:12:10.080 | Why were they moved?

01:12:11.120 | Who ordered that?

01:12:12.640 | What was a classified document doing

01:12:14.280 | in Biden's personal library?

01:12:16.680 | What do the documents pertain to?

01:12:18.480 | Do they touch on the Biden family's business dealings

01:12:22.040 | in Ukraine and China?

01:12:23.760 | So there are real things to look into here.

01:12:25.680 | But let me just take a step back.

01:12:28.880 | Now that the last three presidential candidates

01:12:31.480 | have been ensnared in these classified document problems,

01:12:35.920 | remember, it's Biden now, and then Trump,

01:12:39.440 | and Hillary Clinton before Trump.

01:12:42.040 | I think it's time to step back and ask,

01:12:44.920 | are we over-classifying documents?

01:12:47.280 | I mean, are we fetishizing these documents?

01:12:49.880 | Are they all really that sensitive?

01:12:51.840 | It seems to me that we have an over-classification problem,

01:12:54.760 | meaning that ever since FOIA was passed,

01:12:58.560 | the Freedom of Information Act,

01:13:00.960 | the government can avoid accountability and prying eyes

01:13:04.280 | by simply labeling any document as classified.

01:13:07.720 | So over-classification was a logical response

01:13:10.640 | by the permanent government

01:13:12.240 | to the Freedom of Information Act.

01:13:13.720 | And now it's gotten to the point

01:13:16.160 | where just about everything handed

01:13:18.160 | to a president or vice president is classified.

01:13:20.880 | So I think I can understand

01:13:22.600 | why they're all making this mistake.

01:13:25.320 | And I think a compounding problem

01:13:26.880 | is that we never declassify anything.

01:13:28.480 | There's still all these records

01:13:30.120 | from the Kennedy assassination

01:13:31.800 | that have never been declassified.

01:13:34.720 | - And they're supposed to have declassified these.

01:13:37.280 | The CIA keeps filibustering

01:13:40.640 | on the release of the JFK assassination documents,

01:13:44.640 | and they've been told they have to stop,

01:13:47.120 | and they have to release them,

01:13:47.960 | and then they keep redacting stuff,

01:13:50.560 | which is making it... - They're not releasing them.

01:13:53.000 | - I hate to be a conspiracy theorist here,

01:13:54.680 | but what are they trying to cover up?

01:13:57.040 | I mean, this is a long time ago.

01:13:58.760 | - That's the only way to interpret it.

01:14:00.960 | But even for more mundane documents,

01:14:02.920 | there are very few documents

01:14:04.400 | that need to be classified after even, say, five years.

01:14:07.320 | You could argue that we should be

01:14:08.280 | automatically declassifying them after five years,

01:14:11.160 | unless they go through a process to get reclassified.

01:14:14.280 | I mean, I'd say, like, you guys in business,

01:14:17.800 | I know it's not government, in business,

01:14:19.440 | how many of the documents that you deal with

01:14:22.080 | are still sensitive, are trade secrets, five years later?

01:14:26.400 | - Certainly 20 years later, they're not, right?

01:14:29.440 | Like, in almost all cases.

01:14:30.280 | - No, but I wouldn't even say, like, five years.

01:14:31.280 | I mean, the only documents-- - The Coca-Cola formula.

01:14:33.440 | - The only documents in business that I think I deal with

01:14:36.880 | that you could call sensitive

01:14:38.760 | are the ones that pertain to the company's future plans,

01:14:42.280 | right, because you wouldn't want a competitor--

01:14:43.480 | - Cap table. - To get those.

01:14:45.280 | - Yeah, cap table. - There's a handful of things.

01:14:46.640 | - Legal issues, yeah.

01:14:48.160 | - Even cap table is not that sensitive,

01:14:49.920 | because by the time you go public,

01:14:51.160 | it legally has to be public.

01:14:52.760 | - Yeah, so at some point-- - It's on Carta,

01:14:54.320 | like, there's 100 people who have that.

01:14:55.760 | I mean, it's-- - Exactly.

01:14:56.840 | - So, like, in business, I think our experience has been

01:14:59.520 | there's very few documents that stay sensitive,

01:15:02.840 | that need to remain secret.

01:15:04.800 | Now, look, if Biden or Trump or whoever,

01:15:07.960 | they're reviewing the schematics

01:15:09.960 | to the Javelin missile system or to, you know,

01:15:13.280 | how we make our nuclear bombs or something,

01:15:15.400 | obviously that needs to stay secret forever,

01:15:17.440 | but I don't believe our politicians

01:15:19.080 | are reviewing those kinds of documents.

01:15:21.440 | - Well, I mean, we both--

01:15:22.760 | - I don't really understand what it is

01:15:24.400 | that they're reviewing-- - Why are they keeping them--

01:15:27.040 | - That needs to be classified five years later.

01:15:28.880 | - Well, and why are they keeping them

01:15:29.880 | was the issue we discussed previously.

01:15:31.200 | We actually agreed on that.

01:15:32.640 | I think they're just keeping mementos.

01:15:34.160 | - I think there's a simple explanation

01:15:35.480 | for why they're keeping them, Jason,

01:15:36.920 | which is that everything is more classified,

01:15:41.440 | and there's a zillion documents,

01:15:43.880 | and if you look, like, both Biden and Trump,

01:15:47.160 | these documents were mixed in

01:15:48.360 | with a bunch of personal effects and mementos.

01:15:51.240 | My point is, if you work in government

01:15:53.920 | and handle documents, they're all classified.

01:15:57.200 | So, I mean-- - And if the National Archive

01:15:58.920 | asks for them back, or you find them,

01:16:01.120 | you should just give them back.

01:16:02.160 | I mean, that's gonna wind up being the rug here

01:16:04.400 | is Trump didn't give them back, and Biden did.

01:16:07.440 | So that's the only difference here.

01:16:09.040 | - Well, no, no, no, hold on.

01:16:09.880 | The FBI went to Trump's basement,

01:16:11.320 | they looked around, they said, "Put a lock on this."

01:16:13.280 | They seemed to be okay with it initially.

01:16:15.040 | Then maybe they changed their minds, I don't know.

01:16:17.120 | I'm not defending Trump, but--

01:16:17.960 | - Well, it's pretty clear that he wouldn't give them back in.

01:16:19.880 | That was the issue. - The point I'm making

01:16:20.920 | is that now that Biden, Trump, and Hillary Clinton

01:16:24.320 | have all been ensnared in this,

01:16:26.360 | is it time to rethink the fact

01:16:28.880 | that we're over-classifying so many documents?

01:16:31.760 | I mean, just think about the incentives

01:16:33.360 | that we're creating for our politicians.

01:16:36.440 | Okay, just think about the incentives.

01:16:38.140 | Number one, never use email.

01:16:40.120 | Remember Hillary Clinton and the whole email server?

01:16:42.240 | You gotta be nuts to use email.

01:16:44.060 | Number two, never touch a document.

01:16:46.520 | Never touch a document.

01:16:48.160 | Never let anyone hand you a document.

01:16:50.000 | - Flush them down the toilet.

01:16:51.400 | - Never let anyone hand you a document.

01:16:53.840 | I mean, if you're a politician, an elected official,

01:16:58.840 | the only time you should ever be handling anything

01:17:01.080 | is go into a clean room, make an appointment,

01:17:04.320 | go in there, read something, don't take notes,

01:17:06.680 | don't bring a camera, and then leave.

01:17:08.840 | I mean, this is no way to run a government.

01:17:11.560 | - It's crazy. - Who does this benefit?

01:17:13.280 | Hold on, who does this benefit?

01:17:15.240 | It doesn't benefit our elected officials.

01:17:17.200 | It makes it almost impossible

01:17:18.880 | for them to act like normal people.

01:17:21.000 | - That's why. - It benefits the insiders,

01:17:22.560 | the permanent government.

01:17:23.840 | - You're missing the most important part about the sex.

01:17:26.840 | This was, if you wanna go into conspiracy theories,

01:17:29.960 | this was a setup.

01:17:31.400 | Biden planted the documents

01:17:33.320 | so that we could create the false equivalency

01:17:36.040 | and start up Biden versus Trump 2024.

01:17:38.960 | This ensures that now Trump has something

01:17:41.960 | to fight with Biden about, and this is gonna help Trump.

01:17:45.040 | - 'Cause they're both tainted, equally tainted

01:17:48.280 | from the same source. - Yes, yes.

01:17:49.800 | - They are equally tainted now, but I--

01:17:51.240 | - It puts Trump in the new cycle.

01:17:52.800 | - No, I think it's the opposite.

01:17:54.760 | I think Merrick Garland now is gonna have to drop

01:17:57.800 | the prosecution against Trump for the stolen documents,

01:18:01.120 | or at least that part of what they're investigating him for.

01:18:04.120 | They might still investigate him over January 6th

01:18:05.880 | or something, but they can't investigate Trump

01:18:07.640 | over documents now. - And Georgia.

01:18:08.480 | Those two seems more sticky, yeah.

01:18:10.040 | I agree with that, actually.

01:18:10.880 | I think it's gonna be hard to do.

01:18:12.480 | - But my point is, just think about, look,

01:18:14.640 | both sides are engaged in hyper partisanship.

01:18:16.680 | The way right now that the conservatives on the right,

01:18:20.640 | they're attacking Biden now for the same thing

01:18:23.320 | that the left was attacking Trump for.

01:18:25.160 | My point is, just take a step back,

01:18:27.840 | and again, think about the incentives we're creating

01:18:30.520 | about how to run our government.

01:18:31.920 | You can't use email, and you can't touch documents.

01:18:35.000 | - And everything's an investigation

01:18:37.160 | the second you get on the office.

01:18:38.600 | - And by the way, don't ever go into politics

01:18:41.440 | if you're a business person, because they'll investigate

01:18:43.920 | every deal you ever did prior to getting into politics.

01:18:48.320 | I mean, just think about the incentives we're creating.

01:18:49.160 | - So what are you gonna do when you try to get

01:18:51.400 | your treasury position?

01:18:52.840 | What's gonna happen? - You gotta be nuts.

01:18:53.880 | You gotta be nuts to go into government.

01:18:54.720 | - So you're not gonna take a position

01:18:56.560 | in DeSantis' cabinet? - My point is that

01:18:58.640 | the Washington insiders, by which I mean

01:19:00.280 | the permanent Washington establishment, i.e. the deep state,

01:19:03.720 | they're creating a system in which they're running things,

01:19:07.640 | and the elected officials barely can operate

01:19:09.760 | like normal functioning humans there.

01:19:13.040 | - Interesting. - That's what's going on.

01:19:15.160 | - I heard a great rumor.

01:19:16.720 | This is total gossip mongering.

01:19:18.880 | - Oh, here we go.

01:19:19.720 | - That one of Ken Griffin's best out

01:19:25.920 | is to get DeSantis elected so that he can become

01:19:29.240 | treasury secretary.

01:19:30.320 | I mean, Ken Griffin would get that if he wanted it.

01:19:33.600 | And then he would be able to divest

01:19:35.280 | all of Citadel tax-free.

01:19:37.880 | So he would mark the market like $30 billion,

01:19:40.600 | which is a genius way to go out.

01:19:43.020 | Now, then it occurred to me, oh my God,

01:19:45.520 | that is me and Sachs' path too.

01:19:47.800 | (laughing)

01:19:48.640 | With a lot less money, but the same path.

01:19:51.000 | - Why would it be tax-free?

01:19:54.880 | - When you get appointed to those senior posts,

01:19:58.840 | you're allowed to either stick it in a blind trust,

01:20:01.400 | or you can sell with no capital gains.

01:20:03.640 | - What?

01:20:04.480 | What?

01:20:07.120 | - Well, because they want you to divest.

01:20:09.400 | - They want you unconflicted.

01:20:10.800 | - Yes, anything that presents a conflict,

01:20:12.560 | they want you to divest.

01:20:13.620 | And so the argument is, if you're forced to divest it

01:20:17.100 | to enter a government, you shouldn't be forcibly taxed.

01:20:19.120 | - Wait, if I become mayor of San Francisco or Austin,

01:20:21.400 | I can go to federal government.

01:20:22.240 | - No, no, no, wait, wait, wait.

01:20:23.060 | - It's like having a--

01:20:23.900 | - Secretary of Transportation, J. Cal, you can do that.

01:20:26.360 | - Oh, I'm qualified for that.

01:20:28.320 | I'd take the bus, I got an electric bike.

01:20:30.520 | - To answer Freeberg's point, I think Citadel Securities,

01:20:33.480 | there's a lot of folks that would buy that

01:20:34.960 | because that's just a securities trading business.

01:20:36.760 | And then Citadel, the hedge fund,

01:20:38.720 | probably something like a big bulge bracket bank

01:20:41.000 | for Blackstone, probably Blackstone in fact,

01:20:43.360 | because now Blackstone can plug it into

01:20:45.600 | a trillion dollar asset machine.

01:20:47.600 | I think there would be buyers out the door.

01:20:50.440 | - This is an incredible grift.

01:20:52.360 | Now I know why Sachs--

01:20:53.720 | - It's not a grift at all, but it's an incredible--

01:20:56.120 | - Oh, come on, man, a cabinet position for no cap gains?

01:21:00.040 | - Well, that's not a grift, those are the laws.

01:21:02.800 | They force you to sell everything.

01:21:03.640 | - It so feels grifty to me.

01:21:05.360 | - And then you do public service.

01:21:06.960 | - I think you're misusing the word

01:21:08.560 | to continue to genuflect to the left wing.

01:21:10.600 | - No, I'm not genuflecting.

01:21:11.800 | I think you're being a little defensive

01:21:13.680 | 'cause you see this as a bad--

01:21:14.520 | - Is it genuflect?

01:21:15.360 | - That or you're dumb.

01:21:16.280 | - Big bulge.

01:21:17.120 | (laughing)

01:21:17.960 | - I'm not stupid, man, you don't even know.

01:21:18.800 | - I don't grift when I see it.

01:21:20.320 | You take a cabinet position--

01:21:21.800 | - Sachs, would you take a cabinet position?

01:21:22.640 | - You don't pay cap gains?

01:21:23.600 | - Would you be secretary of the treasury?

01:21:25.000 | - Where does that exist?

01:21:26.520 | - Yeah, Sachs, if you were asked to serve--

01:21:28.320 | - Look, any normal person who wants to serve in government,

01:21:31.400 | you can't use email and you can't touch a document,

01:21:34.360 | and every deal you've ever done gets investigated.

01:21:36.720 | Why not? - That's a yes.

01:21:37.560 | - Why would you wanna do it?

01:21:39.560 | I mean, all of a sudden, you get to divest tax free.

01:21:42.400 | - Methink thou doth protesteth too much, David Sachs.

01:21:45.800 | - The fact that you two know this rule,

01:21:48.080 | and Freebird and I don't.

01:21:50.200 | - No, I know it.

01:21:51.040 | It's like a well-known--

01:21:51.880 | - I'm the only person who doesn't know this.

01:21:53.360 | Rich people know it.

01:21:54.200 | - Jake, I looked up grift.

01:21:55.840 | It means to engage-- - Everyone knows this.

01:21:57.640 | - It means that to engage in a petty

01:21:59.440 | or small-scale swindle.

01:22:01.240 | I don't think selling a $31 billion entity--

01:22:03.400 | - Yeah, no, it's not small.

01:22:04.240 | - To a combination of BlackRock and Blackstone

01:22:06.760 | would be considered a petty, small-scale swindle.

01:22:08.840 | - Did any of you guys watch the Madoff series on Netflix?

01:22:12.720 | - No, was it good? - No.

01:22:14.800 | - Oh my God, it is so depressing.

01:22:17.240 | I gotta say, just that Madoff series,

01:22:19.920 | there is no glimmer of light or hope

01:22:24.360 | or positivity or recourse.

01:22:27.200 | Everyone is a victim.

01:22:28.400 | Everyone suffers.

01:22:29.360 | It is just so dark. - So gross.

01:22:31.080 | - Don't watch it.

01:22:31.920 | It's so depressing.

01:22:32.840 | - The Madoff one.

01:22:34.160 | - The Madoff one is so depressing.

01:22:35.720 | It's so awful. - Yeah, they all kill

01:22:36.800 | themselves and die.

01:22:37.640 | - They become like all the-- - No, one guy died.

01:22:39.080 | - Everyone was a victim. - One guy died of cancer.

01:22:41.960 | - And then Irving Picard, I didn't realize all this,

01:22:43.720 | the trustee that went and got the money,

01:22:45.480 | he went and got money back from these people

01:22:47.400 | who were 80 years old and retired

01:22:48.920 | and had spent that money decades ago,

01:22:50.680 | and he sued them and took their homes away from them,

01:22:53.320 | and no one, and they had no idea--

01:22:55.280 | - No one. - That they were part

01:22:56.160 | of the scam.

01:22:57.000 | No one, no one won.

01:22:58.160 | It was a brutal, awful whole thing, yeah.

01:23:01.000 | - You know what they said.

01:23:02.240 | By the way, that's gonna be really interesting

01:23:03.840 | as we enter this SBF trial because--

01:23:05.480 | - 100%. - 100%.

01:23:07.880 | - That is what happens if you got--

01:23:10.280 | - And that's why the Southern District of New York said

01:23:12.400 | that this case is becoming too big for them

01:23:14.760 | because all the places that SBF sent money,

01:23:17.240 | all those PACs and all those political donations--

01:23:20.160 | - Sent before.

01:23:21.000 | - They have to go and investigate where that money went

01:23:24.000 | and see if they can get it back,

01:23:25.320 | and it's gonna open up an investigation

01:23:26.880 | into each one of these campaign finance

01:23:28.920 | and election and kind of interfering actions

01:23:31.760 | that were taken. - Wait, not Politico.

01:23:32.720 | ProPublica, sorry, ProPublica.

01:23:34.560 | - On the other end of the spectrum,

01:23:36.200 | I did watch this weekend, "Triangle of Sadness."

01:23:39.400 | Have you guys seen this? - I watched it too!

01:23:40.680 | It was great! - Oh my God!

01:23:42.240 | - "Triangle of Sadness" is great!

01:23:43.880 | It's so dark.

01:23:44.880 | - To the Davids, listen, this is one of the,

01:23:48.480 | I thought it was, it didn't pay off the way I thought,

01:23:50.580 | but this is one of the best setups you'll see in a movie.

01:23:53.280 | So basically, it's a bunch of people on a luxury yacht.

01:23:57.800 | So you have a bunch of rich people as the guests.

01:24:00.400 | Then you have the staff that interacts with them.

01:24:04.120 | And this is mostly Caucasian.

01:24:07.280 | And then in the bowels of the ship,

01:24:09.320 | what you see are Asian and black workers

01:24:11.680 | that support them, okay?

01:24:13.160 | So in some ways, it's a little bit

01:24:15.360 | of a microcosm of the world.

01:24:17.160 | - Oh, I thought you were gonna say

01:24:18.240 | a microcosm of something else!

01:24:20.520 | - And then what happens is there's a shipwreck, basically.

01:24:23.920 | - Yeah, oh, don't spoil it, come on.

01:24:25.840 | - Okay, but no, but I'll just tell you.

01:24:28.080 | So the plot is you have this Caucasian patriarchy

01:24:33.080 | that then gets flipped upside down

01:24:36.040 | because after the shipwreck,

01:24:37.200 | the only person who knows how to make a fire

01:24:39.040 | and catch the fish is the Filipino woman

01:24:41.360 | who is in charge of cleaning the toilets.

01:24:43.420 | So she becomes in charge.

01:24:44.700 | So now you flip to this immigrant matriarchy.

01:24:47.440 | - It's a pretty great meditation on class and survival.

01:24:50.880 | It's pretty well done.

01:24:52.200 | - It didn't end well, I thought.

01:24:53.680 | I thought it could.

01:24:54.800 | - Well, it's hard to wrap that one up.

01:24:57.320 | - Well, you know what they say, boys.

01:24:58.600 | Still a little and they throw you in jail.

01:25:00.200 | Still a lot and they make you king.

01:25:02.000 | Famous Bob Dylan quote.

01:25:04.160 | - There you go.

01:25:05.000 | - All right, well, this has been a great episode.

01:25:06.520 | Great to see you, besties.

01:25:08.360 | Austerity menu tonight, Chamath?

01:25:10.280 | What's on the austerity menu tonight?

01:25:12.280 | What are we doing, salad, some tuna sandwiches?

01:25:15.420 | - No, I think Kirsten is doing, I think, durod.

01:25:19.340 | - Durod, yeah, that's, yeah.

01:25:22.320 | - That's a good fish.

01:25:24.800 | - Jake and I once had a great durod in Venice.

01:25:28.440 | - In Venice.

01:25:29.280 | - In Venice.

01:25:30.120 | - The durod from Venice.

01:25:30.940 | - That's one of the best meals we ever had.

01:25:32.360 | Am I right?

01:25:33.200 | - So good, I agree.

01:25:34.120 | - When it's done well, the durod kicks ass.

01:25:36.000 | - There's only one way to cook a durod.

01:25:38.120 | Do you know what that is?

01:25:40.200 | You gotta, it's the way they did in Venice.

01:25:42.560 | You gotta cook the whole fish.

01:25:44.280 | - Yeah, yeah, yeah, she's doing that.

01:25:45.480 | - And then after you cook the fish,

01:25:47.320 | then you de-bone it.

01:25:49.720 | And yeah, that's the way to do it.

01:25:52.000 | - That was back when Saxon and I

01:25:52.960 | used to enjoy each other's company.

01:25:54.320 | (laughing)

01:25:56.560 | - Before this podcast made us into mortal enemies.

01:25:59.280 | - Jake, I'm a little disappointed

01:26:00.800 | you couldn't agree with my take on this documents scandal.

01:26:03.720 | Instead of dunking in a partisan way,

01:26:06.000 | I tried to explain why it was a problem

01:26:08.600 | of our whole political system.

01:26:09.920 | - I like your theory.

01:26:10.760 | I think, you know, you keep making me defend Biden.

01:26:15.120 | I think Biden's a grifter.

01:26:16.840 | I told you these guys are grifting.

01:26:19.160 | Just think you're, I just think your party

01:26:20.840 | grifts a little bit more.

01:26:23.480 | But yeah, compare your grift.

01:26:24.880 | - Are we going to play Saturday after the wild card game?

01:26:27.000 | Are you guys interested in playing Saturday as well?

01:26:28.600 | 'Cause I got the hall pass.

01:26:29.560 | I can do a game on Saturday.

01:26:32.000 | - I don't know.

01:26:32.840 | I have to check with my boss.

01:26:34.280 | - Who's going to the, are you guys all going?

01:26:36.520 | Sax, are you going to come to play poker

01:26:38.440 | at that live stream thing for the day, LA?

01:26:42.040 | - I doubt it, no.

01:26:43.160 | - He doesn't want to interact with humans.

01:26:46.040 | - That does not play well in confirmation hearings.

01:26:49.480 | - No.

01:26:50.320 | - The last time I did one of those,

01:26:51.440 | Alan Keating destroyed me on camera.

01:26:53.560 | I like, I had like, I had two left feet

01:26:56.920 | and every time he bluffed, I folded.

01:26:59.400 | Every time he had the nuts, I called.

01:27:01.480 | It was brutal.

01:27:02.320 | - That's true.

01:27:03.160 | - Now you do this.

01:27:04.000 | - That was a bad one.

01:27:04.840 | - A shellacking, a classic Sax shellacking.

01:27:07.680 | - So it's an edge saving thing.

01:27:08.680 | Is that what's going on here?

01:27:09.640 | - No, no, no, no.

01:27:10.480 | - Preserving.

01:27:11.320 | - No, I just said I don't have time to do it.

01:27:12.160 | - No, it has to do with the cabinet positions.

01:27:13.680 | He doesn't need to be seen recklessly gambling.

01:27:16.480 | It's a bad look.

01:27:17.320 | - If you could pick any cabinet position, Sax,

01:27:19.120 | which one would it be?

01:27:20.320 | - State.

01:27:21.160 | - Treasury.

01:27:23.120 | - State's a lot of travel.

01:27:24.120 | - Say the word, Sax.

01:27:24.960 | - State's a lot of travel.

01:27:25.800 | You never stay at home.

01:27:26.620 | You're always on a plane.

01:27:27.460 | - I'll tell you, I don't know.

01:27:28.300 | - That's what he's looking for.

01:27:30.400 | - I don't know that those like cabinet positions

01:27:32.800 | are that important.

01:27:33.680 | I mean, they run these giant bureaucracies

01:27:36.440 | that again are permanent.

01:27:37.920 | You can't fire anyone.

01:27:39.320 | So if you can't fire a person,

01:27:41.120 | do they really report to you?

01:27:42.600 | - Right.

01:27:43.440 | - I mean, only ceremonial.

01:27:44.640 | - Trump's idea was like, put a bunch of hardline,

01:27:47.240 | CEO type people in charge,

01:27:49.000 | have them blow up these things and make it more efficient.

01:27:51.040 | It didn't really work, did it?

01:27:52.560 | - Yeah, well, you know why a CEO is actually in charge?

01:27:55.680 | Like Elon, he walks in,

01:27:57.400 | if he doesn't like what you're doing, he'll just fire you.

01:27:59.640 | You can't fire anyone.

01:28:00.720 | How do you manage them

01:28:01.840 | when they don't have to listen to anything you say?

01:28:04.600 | That's our whole government right now.

01:28:06.500 | Our cabinet heads are figureheads

01:28:11.000 | for these departments, these giant departments.

01:28:13.000 | - Is that a no, or is that a yes, you'd still take state?

01:28:16.240 | Look at that.

01:28:18.760 | - Well, I think I have another theory.

01:28:20.840 | I think he's going for the ambassadorship first.

01:28:23.240 | What is the best ambassadorship?

01:28:24.960 | - Well, you can't divest everything with no cash.

01:28:27.520 | - Historically, you can tell which ambassadorship

01:28:30.440 | is the best one based on how much they charge for it.

01:28:32.720 | - Yeah.

01:28:33.560 | - So I think London is the most expensive.

01:28:36.040 | I think that one's $15 million.

01:28:36.880 | - It's 10 million for London, 10 to 15.

01:28:38.960 | - 10, 15 million, yeah.

01:28:40.760 | - 10, 15 million?

01:28:42.640 | That's what Sax is fourth least expensive home cost.

01:28:47.280 | - No, no, no, you have to spend that every year

01:28:48.680 | to run it, Jason.

01:28:50.040 | You only get, you get a fixed-

01:28:50.880 | - That's nothing for him.

01:28:51.840 | - You could be the ambassador to Guinea

01:28:53.880 | or the ambassador to the UK, you get the same budget.

01:28:56.800 | - Actually, what's kind of funny is I know two people

01:28:58.680 | who served as ambassadors under Trump

01:29:00.560 | and it was really cheap to get those

01:29:02.080 | because no one wanted to be part

01:29:04.520 | of the Trump administration.

01:29:06.480 | - Oh, they were on fire sale, two for one.

01:29:08.400 | - They were on fire sale after, because of Trump.

01:29:12.080 | - Who wants to be tainted?

01:29:13.680 | - But by the way, one of them,

01:29:14.720 | and you can just bleep out the name,

01:29:16.560 | (beep)

01:29:17.560 | was telling me it was the best thing

01:29:19.040 | because he ended up selling,

01:29:20.160 | they already got the all-time highs to take the job.

01:29:22.480 | He was like, "I got to get out of all of this stuff."

01:29:24.240 | - No, but listen, let me tell you, the ambassadorships,

01:29:26.920 | it was a smart trade by those guys

01:29:28.560 | because ambassador's a lifetime title.

01:29:31.040 | So you're ambassador, whatever.

01:29:33.400 | No one remembers what president when you were ambassador,

01:29:35.680 | no one cares.

01:29:36.520 | - So you are going for the ambassador.

01:29:37.760 | - No, no. - So Steve,

01:29:38.600 | I think it's fair to say.

01:29:39.440 | - I'm not, I'm not interested in ceremonial things.

01:29:43.840 | I'm interested in making an impact.

01:29:46.120 | And the problem with all these positions,

01:29:48.400 | I mean, being a cabinet official

01:29:51.120 | is not much different than being an ambassador.

01:29:53.480 | - So you're going to enlist in the Navy?

01:29:56.680 | - No.

01:29:57.520 | What has a bigger impact?

01:29:59.880 | Being on all in pod or being an ambassador?

01:30:02.760 | Who's more influential?

01:30:04.200 | Sax on the all in pod or beep as the ambassador of Sweden?

01:30:08.480 | - Being on the all in pod actually.

01:30:10.000 | - All in pod is more impactful.

01:30:13.280 | - By the way, this is why I take issue

01:30:14.760 | with your statement about the term mainstream media.

01:30:17.040 | 'Cause I think you have become the mainstream media

01:30:19.600 | more than most of the folks that-

01:30:20.920 | - No, we're independent media.

01:30:22.320 | We're independent media.

01:30:23.280 | - Trust me, it's independent.

01:30:24.320 | This thing's hanging on by a thread.

01:30:26.240 | - And stop genuflecting.

01:30:27.920 | (laughing)

01:30:29.200 | - No, it's independent.

01:30:30.240 | Who knows if this thing's going to last

01:30:32.640 | another three episodes.

01:30:34.120 | - I just like saying the word genuflect.

01:30:35.840 | - You like genuflecting, I know.

01:30:37.440 | - That is the top word of 2023 so far for me.

01:30:40.360 | - Oh, is that, is somebody doing an analysis

01:30:42.640 | with Chad Chappity of the words used here?

01:30:44.560 | - No, but Saks brought that word up.

01:30:45.840 | It was just, it's a wonderful word.

01:30:47.520 | It's not used enough.

01:30:48.720 | All right, everybody, we'll see you next time

01:30:51.920 | on the all in podcast.

01:30:53.080 | Comments are turned back on.

01:30:54.920 | Have at it, you animals.

01:30:56.520 | Love you guys.

01:30:57.360 | - Enjoy, bye-bye. - Love you besties, bye.

01:30:59.000 | (upbeat music)

01:31:02.600 | (upbeat music)

01:31:05.200 | (upbeat music)

01:31:08.040 | (upbeat music)

01:31:10.640 | (upbeat music)

01:31:15.720 | (upbeat music)

01:31:23.000 | (upbeat music)

01:31:28.440 | (upbeat music)

01:31:33.520 | (upbeat music)

01:31:36.120 | (upbeat music)

01:31:41.520 | (upbeat music)

01:31:48.120 | (upbeat music)

01:31:56.120 | (eerie music)

01:31:56.620 | <3

01:31:57.120 | (Music plays)

E111: Microsoft to invest $10B in OpenAI, generative AI hype, America's over-classification problem

Chapters