GPT 4 - hype vs reality
00:00:00.000 | You may have heard rumours that ChatGPT 4 is going to be released imminently, say in January,
00:00:06.580 | and that it's going to dwarf ChatGPT 3's ability, just like you can see in this graph.
00:00:12.480 | Well, I'm going to let Sam Altman correct the record here on both fronts.
00:00:17.140 | And then at the end of the video, I'm going to discuss what I think ChatGPT 4 will be capable of.
00:00:22.140 | Here's Sam Altman on the timing of the release of ChatGPT 4.
00:00:27.100 | Can you comment on whether GPT 4 is coming out in the first quarter, first half of the year?
00:00:32.440 | It'll come out at some point when we are like confident that we can do it safely and responsibly.
00:00:38.640 | I think in general, we are going to release technology much more slowly than people would like.
00:00:45.840 | We're going to sit on it for much longer than people would like.
00:00:48.300 | Then referring to ChatGPT 4, he talked about whether it will be an exponential increase in terms of its abilities
00:00:55.800 | or more of an incremental upgrade.
00:00:58.040 | Given the magnitude of the economic impact we expect here, more gradual is better.
00:01:03.700 | And so putting out a very weak and imperfect system like ChatGPT and then making it a little better this year,
00:01:09.400 | a little better later this year, a little better next year, that seems much better than the alternative.
00:01:14.040 | Then Sam directly addressed the hype train that had been generated by graphics such as this one.
00:01:21.000 | I saw a visual, and I don't know if it was accurate, but it showed GPT 3.5 versus, I guess, what GPT 4 is expected.
00:01:28.040 | I saw that thing on Twitter.
00:01:29.240 | Did you? Was that accurate?
00:01:30.740 | Complete bullshit. No.
00:01:31.500 | Okay, because that was a little bit scary.
00:01:33.940 | The GPT 4 rumor mill is like a ridiculous thing.
00:01:40.240 | I don't know where it all comes from.
00:01:41.740 | I don't know why people don't have better things to speculate on.
00:01:44.440 | I get a little bit of it, like it's sort of fun, but that it's been going for like six months at this volume.
00:01:50.200 | People are begging to be disappointed, and they will be.
00:01:52.940 | Like it's, you know, people are going to like, the hype is just like, we don't have an actual AGI.
00:01:58.060 | And I think that's sort of what is expected of us.
00:02:00.000 | And, you know, yeah, we're going to disappoint those people.
00:02:02.440 | In a moment, you're going to see Sam respond to a question about ChatGPT putting Google out of business.
00:02:09.260 | And he made a fantastic point, which I'm going to go into more detail on after he gives his remarks.
00:02:16.200 | Yeah, I think whenever someone like talks about a technology being the end of some other giant company, it's usually wrong.
00:02:21.500 | Like I think people forget they get to make a counter move here and they're like pretty smart, pretty competent.
00:02:27.160 | The counter move that Sam is referring to may well be Palm, which is a 540 billion parameter transformer model.
00:02:37.200 | And you can see from the graphic the improvements that have been made by increasing the number of parameters.
00:02:43.900 | But I've done some research on Palm, see this graph, notice the line crossing the performance of the average human.
00:02:52.400 | It admits when you go into detail about the tasks that Palm can achieve,
00:02:57.100 | that we're still talking about solving 60% of problems that 9 to 12 year olds can solve.
00:03:04.600 | And that is incredible.
00:03:05.960 | So the average 12 year old can solve 60% of them and Palm can solve 58% of them.
00:03:11.440 | But that isn't exactly AGI, not yet, at least.
00:03:14.640 | So if ChatGPT-4 is anywhere in that kind of ballpark, we're going to notice the difference.
00:03:20.340 | We're going to see the improvement just as that graph increased quite dramatically by increasing the number of parameters.
00:03:26.600 | Maybe it will be able to explain jokes, for example, understand books a bit better.
00:03:31.300 | But we're definitely not talking about AGI, not yet.
00:03:35.400 | I am certain that the initial buzz around ChatGPT-4 is going to be amazing.
00:03:40.060 | But I'm going to let Sam address the difference between the initial buzz,
00:03:44.500 | the impressive achievements it can do, and the robustness of the entire model.
00:03:49.500 | One of the sort of strange things about these technologies is they are impressive, but not robust.
00:03:55.100 | And so you use them in a first demo, you kind of have this like very impressive, like, wow, this is like incredible and ready to go.
00:04:02.860 | You use them 100 times, you see the weaknesses.
00:04:05.460 | And so I think people can get a much sort of a false impression of how good they are.
00:04:12.540 | However, that's all going to get better.
00:04:15.100 | The critics who point these problems out and say, well, this is why it's like, you know, all like, like, you know, fake news or whatever, are equally wrong.
00:04:22.740 | Thank you for watching and do subscribe for more such content.