Sophia is not AGI (Ben Goertzel) | AI Podcast Clips with Lex Fridman
Chapters
0:00:20 David Hanson
7:7 Agi Ethics
22:25 How Sofia Works
00:00:00.000 | You're the chief scientist of Hanson Robotics.
00:00:05.920 | You're still involved with Hanson Robotics, doing a lot of really interesting stuff there.
00:00:10.620 | This is, for people who don't know, the company that created Sophia the Robot.
00:00:15.320 | Can you tell me who Sophia is?
00:00:19.200 | I'd rather start by telling you who David Hanson is.
00:00:23.120 | David is the brilliant mind behind the Sophia Robot.
00:00:28.080 | So far he remains more interesting than his creation, although she may be improving faster
00:00:34.000 | than he is, actually.
00:00:35.000 | That's a good point.
00:00:36.000 | I met David maybe 2007 or something at some futurist conference we were both speaking
00:00:45.480 | at, and I could see we had a great deal in common.
00:00:50.600 | I mean, we were both kind of crazy, but we also both had a passion for AGI and the singularity,
00:00:59.400 | and we were both huge fans of the work of Philip K. Dick, the science fiction writer.
00:01:05.280 | I wanted to create benevolent AGI that would create massively better life for all humans
00:01:13.320 | and all sentient beings, including animals, plants, and superhuman beings.
00:01:19.160 | And he wanted exactly the same thing, but he had a different idea of how to do it.
00:01:24.200 | He wanted to get computational compassion.
00:01:28.120 | He wanted to get machines that would love people and empathize with people, and he thought
00:01:34.040 | the way to do that was to make a machine that could look people eye to eye, face to face,
00:01:40.460 | look at people and make people love the machine, and the machine loves the people back.
00:01:45.400 | So I thought that was a very different way of looking at it, because I'm very math-oriented,
00:01:50.200 | and I'm just thinking, like, what is the abstract cognitive algorithm that will let the system
00:01:56.920 | internalize the complex patterns of human values, blah, blah, blah, whereas he's like,
00:02:02.600 | look you in the face and the eye and love you, right?
00:02:05.240 | So we hit it off quite well, and we talked to each other off and on.
00:02:12.400 | Then I moved to Hong Kong in 2011.
00:02:17.280 | So I've been living all over the place.
00:02:21.120 | I've been in Australia and New Zealand in my academic career, then in Las Vegas for
00:02:26.160 | a while, was in New York in the late '90s starting my entrepreneurial career, was in
00:02:31.720 | DC for nine years doing a bunch of US government consulting stuff, then moved to Hong Kong
00:02:37.280 | in 2011, mostly because I met a Chinese girl who I fell in love with, and we got married.
00:02:43.960 | She's actually not from Hong Kong, she's from mainland China, but we converged together
00:02:47.800 | in Hong Kong, still married now, have a two-year-old baby.
00:02:52.040 | So went to Hong Kong to see about a girl, I guess.
00:02:55.320 | Yeah, pretty much, yeah.
00:02:57.040 | And on the other hand, I started doing some cool research there with Jin You at Hong Kong
00:03:02.040 | Polytechnic University.
00:03:04.280 | I got involved with a project called IDEA using machine learning for stock and futures
00:03:08.400 | prediction, which was quite interesting.
00:03:11.200 | And I also got to know something about the consumer electronics and hardware manufacturer
00:03:16.320 | ecosystem in Shenzhen, across the border, which is the only place in the world that
00:03:21.080 | makes sense to make complex consumer electronics at large scale and low cost.
00:03:26.600 | It's astounding, the hardware ecosystem that you have in South China.
00:03:32.960 | People here cannot imagine what it's like.
00:03:35.120 | So David was starting to explore that also.
00:03:39.800 | I invited him to Hong Kong to give a talk at Hong Kong PolyU, and I introduced him in
00:03:45.200 | Hong Kong to some investors who were interested in his robots.
00:03:49.480 | And he didn't have Sophia then, he had a robot of Philip K. Dick, our favorite science fiction
00:03:54.240 | writer.
00:03:55.240 | He had a robot Einstein, he had some little toy robots that looked like his son Zeno.
00:03:59.880 | So through the investors I connected him to, he managed to get some funding to basically
00:04:05.680 | port Hanson Robotics to Hong Kong.
00:04:08.640 | And when he first moved to Hong Kong, I was working on AGI research and also on this machine
00:04:14.980 | learning trading project.
00:04:17.200 | So I didn't get that tightly involved with Hanson Robotics.
00:04:20.960 | But as I hung out with David more and more, as we were both there in the same place, I
00:04:27.280 | started to think about what you could do to make his robots smarter than they were.
00:04:36.440 | And so we started working together.
00:04:38.200 | And for a few years I was chief scientist and head of software at Hanson Robotics.
00:04:43.760 | Then when I got deeply into the blockchain side of things, I stepped back from that and
00:04:49.700 | co-founded SingularityNet.
00:04:52.340 | David Hanson was also one of the co-founders of SingularityNet.
00:04:55.620 | So part of our goal there had been to make the blockchain-based CloudMind platform for
00:05:02.680 | Sophia and the other--
00:05:04.240 | So Sophia would be just one of the robots in SingularityNet.
00:05:08.880 | Yeah, yeah, yeah, exactly.
00:05:12.720 | Many copies of the Sophia robot would be among the user interfaces to the globally distributed
00:05:20.600 | SingularityNet CloudMind.
00:05:22.320 | And David and I talked about that for quite a while before co-founding SingularityNet.
00:05:28.480 | By the way, in his vision and your vision, was Sophia tightly coupled to a particular
00:05:36.000 | AI system?
00:05:37.000 | Or was the idea that you can just keep plugging in different AI systems within the head of
00:05:42.760 | it?
00:05:43.760 | I think David's view was always that Sophia would be a platform, much like, say, the Pepper
00:05:51.640 | robot is a platform from SoftBank.
00:05:54.760 | Should be a platform with a set of nicely designed APIs that anyone can use to experiment
00:06:01.280 | with their different AI algorithms on that platform.
00:06:06.640 | And SingularityNet, of course, fits right into that, right?
00:06:09.440 | Because SingularityNet, it's an API marketplace.
00:06:11.940 | So anyone can put their AI on there.
00:06:14.320 | OpenCog is a little bit different.
00:06:16.840 | I mean, David likes it, but I'd say it's my thing.
00:06:20.040 | It's not his.
00:06:21.040 | I think David has a little more passion for biologically-based approaches to AI than I
00:06:26.400 | do, which makes sense.
00:06:27.880 | I mean, he's really into human physiology and biology.
00:06:30.760 | He's a character sculptor, right?
00:06:35.760 | But he also worked a lot with rule-based and logic-based AI systems, too.
00:06:39.520 | So, yeah, he's interested in not just Sophia, but all the handsome robots as a powerful
00:06:45.120 | social and emotional robotics platform.
00:06:49.280 | And what I saw in Sophia was a way to get AI algorithms out there in front of a whole
00:07:01.400 | lot of different people in an emotionally compelling way.
00:07:04.320 | And part of my thought was really kind of abstract, connected to AGI ethics.
00:07:09.520 | And many people are concerned AGI is going to enslave everybody or turn everybody into
00:07:16.560 | computronium to make extra hard drives for their cognitive engine or whatever.
00:07:23.560 | And emotionally, I'm not driven to that sort of paranoia.
00:07:29.400 | I'm really just an optimist by nature.
00:07:32.040 | But intellectually, I have to assign a non-zero probability to those sorts of nasty outcomes.
00:07:40.040 | Because if you're making something 10 times as smart as you, how can you know what it's
00:07:43.600 | going to do?
00:07:44.600 | You've got to do some uncertainty there, just as my dog can't predict what I'm going
00:07:49.120 | to do tomorrow.
00:07:50.680 | So it seemed to me that based on our current state of knowledge, the best way to bias the
00:07:58.960 | AGIs we create toward benevolence would be to infuse them with love and compassion the
00:08:07.800 | way that we do our own children.
00:08:09.540 | So you want to interact with AIs in the context of doing compassionate, loving, and beneficial
00:08:16.240 | things.
00:08:17.880 | And in that way, as your children will learn by doing compassionate, beneficial, loving
00:08:21.740 | things alongside you, and that way the AI will learn in practice what it means to be
00:08:27.340 | compassionate, beneficial, and loving, it will get a sort of ingrained intuitive sense
00:08:33.660 | of this, which it can then abstract in its own way as it gets more and more intelligent.
00:08:38.620 | Now David saw this the same way.
00:08:40.700 | That's why he came up with the name Sophia, which means wisdom.
00:08:46.060 | So it seemed to me making these beautiful, loving robots to be rolled out for beneficial
00:08:52.020 | applications would be the perfect way to roll out early stage AGI systems so they can learn
00:09:00.620 | from people, and not just learn factual knowledge, but learn human values and ethics from people
00:09:06.280 | while being their home service robots, their education assistants, their nursing robots.
00:09:11.900 | So that was the grand vision.
00:09:13.940 | Now, if you've ever worked with robots, the reality is quite different, right?
00:09:18.380 | The first principle is the robot is always broken.
00:09:22.900 | I mean, I worked with robots in the 90s a bunch, when you had to solder them together
00:09:26.900 | yourself.
00:09:27.900 | And I'd put neural nets doing reinforcement learning on overturned salad bowl type robots
00:09:34.740 | in the 90s when I was a professor.
00:09:37.300 | Things of course advanced a lot, but...
00:09:39.900 | But the principle still holds.
00:09:40.900 | Yeah, the principle of the robot's always broken still holds.
00:09:43.940 | Yeah.
00:09:44.940 | So faced with the reality of making Sophia do stuff, many of my robo-AGI aspirations
00:09:51.900 | were temporarily cast aside.
00:09:54.420 | And I mean, there's just a practical problem of making this robot interact in a meaningful
00:10:01.100 | way, because you put nice computer vision on there, but there's always glare.
00:10:06.140 | And then you have a dialogue system, but at the time I was there, no speech-to-text algorithm
00:10:14.060 | could deal with Hong Kongese people's English accents.
00:10:17.700 | So the speech-to-text was always bad.
00:10:19.500 | So the robot always sounded stupid, because it wasn't getting the right text, right?
00:10:23.580 | So I started to view that really as what in software engineering you call a walking skeleton,
00:10:30.780 | which is maybe the wrong metaphor to use for Sophia, or maybe the right one.
00:10:34.300 | I mean, where the walking skeleton is in software development is, if you're building a complex
00:10:39.900 | system, how do you get started?
00:10:41.780 | Well, one way is to first build part one well, then build part two well, then build part
00:10:45.540 | three well, and so on.
00:10:47.220 | And the other way is you make a simple version of the whole system and put something in the
00:10:52.220 | place of every part the whole system will need, so that you have a whole system that
00:10:56.200 | does something, and then you work on improving each part in the context of that whole integrated
00:11:02.020 | system.
00:11:03.020 | So that's what we did on a software level in Sophia.
00:11:05.980 | We made like a walking skeleton software system where, so there's something that sees, there's
00:11:10.940 | something that hears, there's something that moves, there's something that remembers, there's
00:11:16.060 | something that learns.
00:11:17.840 | You put a simple version of each thing in there, and you connect them all together so
00:11:22.420 | that the system will do its thing.
00:11:24.540 | So there's a lot of AI in there.
00:11:27.600 | There's not any AGI in there.
00:11:29.140 | I mean, there's computer vision to recognize people's faces, recognize when someone comes
00:11:33.840 | in the room and leaves, try to recognize whether two people are together or not.
00:11:38.540 | I mean, the dialogue system, it's a mix of like hand-coded rules with deep neural nets
00:11:46.300 | that come up with their own responses.
00:11:49.580 | And there's some attempt to have a narrative structure and sort of try to pull the conversation
00:11:56.200 | into something with a beginning, middle, and end in this sort of story arc.
00:12:01.300 | I mean, like if you look at the Lobner Prize and the systems that beat the Turing test
00:12:06.500 | currently, they're heavily rule-based because like you said, narrative structure to create
00:12:12.060 | compelling conversations, you currently, neural networks cannot do that well, even with Google
00:12:16.860 | Amina, when you actually look at full-scale conversations, it's just not--
00:12:21.300 | Yeah, this is the thing.
00:12:22.300 | So I've actually been running an experiment the last couple of weeks taking Sophia's chat
00:12:28.260 | bot and then Facebook's transformer chat bot, which they opened the model.
00:12:33.220 | We've had them chatting to each other for a number of weeks on the server.
00:12:36.820 | That's funny.
00:12:37.900 | We're generating training data of what Sophia says in a wide variety of conversations.
00:12:43.420 | But we can see compared to Sophia's current chat bot, the Facebook deep neural chat bot
00:12:50.900 | comes up with a wider variety of fluent sounding sentences.
00:12:55.200 | On the other hand, it rambles like mad.
00:12:57.980 | The Sophia chat bot, it's a little more repetitive in the sentence structures it uses.
00:13:04.540 | On the other hand, it's able to keep like a conversation arc over a much longer period.
00:13:11.020 | Now you can probably surmount that using Reformer and using various other deep neural architectures
00:13:18.980 | to improve the way these transformer models are trained.
00:13:21.980 | But in the end, neither one of them really understands what's going on.
00:13:26.820 | And I mean, that's the challenge I had with Sophia is if I were doing a robotics project
00:13:34.260 | aimed at AGI, I would want to make like a robo toddler that was just learning about
00:13:38.820 | what it was seeing because then the language is grounded in the experience of the robot.
00:13:42.800 | But what Sophia needs to do to be Sophia is talk about sports or the weather or robotics
00:13:49.060 | or the conference she's talking at.
00:13:51.940 | She needs to be fluent talking about any damn thing in the world.
00:13:56.420 | And she doesn't have grounding for all those things.
00:14:00.340 | So there's this, just like, I mean, Google MENA and Facebook's chat bot don't have grounding
00:14:05.020 | for what they're talking about either.
00:14:07.920 | So in a way, the need to speak fluently about things where there's no non-linguistic grounding
00:14:15.620 | pushes what you can do for Sophia in the short term a bit away from AGI.
00:14:24.020 | - I mean, it pushes you towards IBM Watson situation where you basically have to do heuristic
00:14:30.460 | and hard code stuff and rule-based stuff.
00:14:32.820 | I have to ask you about this.
00:14:35.820 | So because, in part, Sophia is an art creation because it's beautiful.
00:14:49.380 | She's beautiful because she inspires through our human nature of anthropomorphize things.
00:14:57.460 | We immediately see an intelligent being there.
00:15:00.580 | - Because David is a great sculptor.
00:15:01.980 | - Is a great sculptor, that's right.
00:15:03.380 | So in fact, if Sophia just had nothing inside her head, said nothing, if she just sat there,
00:15:11.220 | we're already prescribed some intelligence to--
00:15:13.660 | - There's a long selfie line in front of her after every talk.
00:15:16.740 | - That's right.
00:15:17.900 | So it captivated the imagination of many people.
00:15:21.620 | I wasn't gonna say the world, but yeah, I mean, a lot of people.
00:15:25.180 | - Billions of people, which is amazing.
00:15:28.140 | - It's amazing, right.
00:15:29.940 | Now, of course, many people have prescribed essentially AGI type of capabilities to Sophia
00:15:38.820 | when they see her.
00:15:40.700 | And of course, friendly French folk like Yann LeCun immediately see that, the people from
00:15:49.740 | the AI community, and get really frustrated because--
00:15:53.060 | - It's understandable.
00:15:56.100 | - And then they criticize people like you who sit back and don't say anything about,
00:16:04.100 | like basically allow the imagination of the world, allow the world to continue being captivated.
00:16:11.780 | So what's your sense of that kind of annoyance that the AI community has?
00:16:18.660 | - Well, I think there's several parts to my reaction there.
00:16:23.440 | First of all, if I weren't involved with Hanson & Box and didn't know David Hanson personally,
00:16:31.180 | I probably would have been very annoyed initially at Sophia as well.
00:16:35.780 | I mean, I can understand the reaction.
00:16:37.260 | I would have been like, wait, all these stupid people out there think this is an AGI, but
00:16:44.300 | it's not an AGI, but they're tricking people that this very cool robot is an AGI.
00:16:50.900 | And now those of us trying to raise funding to build AGI, people will think it's already
00:16:56.980 | there and already works, right?
00:16:59.100 | So on the other hand, I think even if I weren't directly involved with it, once I dug a little
00:17:07.000 | deeper into David and the robot and the intentions behind it, I think I would have stopped being
00:17:14.100 | pissed off, whereas folks like Jan LeCun have remained pissed off after their initial reaction.
00:17:22.380 | - That's his thing.
00:17:23.380 | - I think that in particular struck me as somewhat ironic because Jan LeCun is working
00:17:31.660 | for Facebook, which is using machine learning to program the brains of the people in the
00:17:37.580 | world toward vapid consumerism and political extremism.
00:17:42.760 | So if your ethics allows you to use machine learning in such a blatantly destructive way,
00:17:51.420 | why would your ethics not allow you to use machine learning to make a lovable theatrical
00:17:56.980 | robot that draws some foolish people into its theatrical illusion?
00:18:04.140 | If the pushback had come from Yoshua Bengio, I would have felt much more humbled by it
00:18:08.700 | because he's not using AI for blatant evil, right?
00:18:13.340 | On the other hand, he also is a super nice guy and doesn't bother to go out there trashing
00:18:19.060 | other people's work for no good reason.
00:18:22.540 | - Shots fired.
00:18:23.540 | But I get you.
00:18:24.540 | - If you're gonna ask, I'm gonna answer.
00:18:28.700 | - No, for sure.
00:18:29.700 | I think we'll go back and forth.
00:18:31.100 | I'll talk to Jan again.
00:18:32.300 | - I would add on this, though.
00:18:36.020 | David Hansen is an artist and he often speaks off the cuff.
00:18:42.140 | And I have not agreed with everything that David has said or done regarding Sophia.
00:18:47.460 | And David also does not agree with everything David has said or done about Sophia.
00:18:53.540 | I mean, David is an artistic wild man and that's part of his charm.
00:19:01.580 | That's part of his genius.
00:19:02.580 | So certainly there have been conversations within Hansen Robotics and between me and
00:19:09.500 | David where I was like, "Let's be more open about how this thing is working."
00:19:16.140 | And I did have some influence in nudging Hansen Robotics to be more open about how Sophia
00:19:23.780 | was working.
00:19:25.540 | And David wasn't especially opposed to this.
00:19:28.500 | And he was actually quite right about it.
00:19:30.300 | What he said was, "You can tell people exactly how it's working and they won't care.
00:19:35.860 | They want to be drawn into the illusion."
00:19:37.500 | And he was 100% correct.
00:19:40.340 | I'll tell you what, this wasn't Sophia, this was Philip K. Dick.
00:19:43.660 | But we did some interactions between humans and Philip K. Dick Robot in Austin, Texas
00:19:50.620 | a few years back.
00:19:51.780 | And in this case, the Philip K. Dick was just teleoperated by another human in the other
00:19:55.620 | room.
00:19:56.620 | So during the conversations, we didn't tell people the robot was teleoperated.
00:20:00.700 | We just said, "Here, have a conversation with Phil Dick.
00:20:03.340 | We're going to film you."
00:20:05.020 | And they had a great conversation with Philip K. Dick, teleoperated by my friend, Stefan
00:20:10.260 | Bugaj.
00:20:11.260 | After the conversation, we brought the people in the back room to see Stefan, who was controlling
00:20:18.700 | the Philip K. Dick Robot, but they didn't believe it.
00:20:22.660 | These people were like, "Well, yeah, but I know I was talking to Phil.
00:20:26.100 | Like, maybe Stefan was typing, but the spirit of Phil was animating his mind while he was
00:20:31.940 | typing."
00:20:32.940 | So even though they knew it was a human in the loop, even seeing the guy there, they
00:20:37.420 | still believed that was Phil they were talking to.
00:20:40.420 | A small part of me believes that they were right, actually.
00:20:44.740 | Because our understanding...
00:20:45.740 | Well, we don't understand the universe.
00:20:46.740 | That's the thing.
00:20:47.740 | I mean, there is a cosmic mind field that we're all embedded in that yields many strange
00:20:53.820 | synchronicities in the world, which is a topic we don't have time to go into too much here.
00:21:00.380 | There's something to this where our imagination about Sophia and people like Jan LeCun being
00:21:09.740 | frustrated about it is all part of this beautiful dance of creating artificial intelligence
00:21:15.140 | that's almost essential.
00:21:16.840 | You see with Boston Dynamics, whom I'm a huge fan of as well, you know, the kind of...
00:21:22.060 | I mean, these robots are very far from intelligent.
00:21:25.580 | I played with their last one, actually.
00:21:29.940 | With the spot mini.
00:21:30.940 | Yeah, very cool.
00:21:31.940 | I mean, it reacts quite in a fluid and flexible way.
00:21:34.940 | But we immediately ascribe the kind of intelligence, we immediately ascribe AGI to them.
00:21:40.300 | Yeah, yeah, if you kick it and it falls down and goes, "Ow," you feel bad, right?
00:21:43.540 | You can't help it.
00:21:45.380 | And I mean, that's going to be part of our journey in creating intelligent systems.
00:21:52.500 | More and more and more and more.
00:21:54.380 | As Sophia starts out with a walking skeleton, as you add more and more intelligence, I mean,
00:22:00.180 | we're going to have to deal with this kind of idea.
00:22:02.380 | Absolutely.
00:22:03.380 | And about Sophia, I would say, first of all, I have nothing against Jan LeCun.
00:22:07.540 | No, no, this is fine.
00:22:08.540 | He's a nice guy.
00:22:10.420 | If he wants to play the media banter game, I'm happy to play it.
00:22:15.940 | He's a good researcher and a good human being.
00:22:19.620 | I'd happily work with the guy.
00:22:21.620 | The other thing I was going to say is, I have been explicit about how Sophia works.
00:22:28.220 | And I've posted online, on H+ Magazine, an online webzine, I mean, I posted a moderately
00:22:36.540 | detailed article explaining, like, there are three software systems we've used inside Sophia.
00:22:43.220 | There's a timeline editor, which is like a rule-based authoring system, where she's really
00:22:47.100 | just being an outlet for what a human scripted.
00:22:50.620 | There's a chatbot, which has some rule-based and some neural aspects.
00:22:54.380 | And then sometimes we've used OpenCog behind Sophia, where there's more learning and reasoning.
00:22:59.980 | And you know, the funny thing is, I can't always tell which system is operating here,
00:23:05.220 | right?
00:23:06.220 | Because whether she's really learning or thinking, or just appears to be, over a half
00:23:11.580 | hour I could tell, but over like three or four minutes of interaction, I couldn't tell.
00:23:16.180 | So even having three systems that's already sufficiently complex, where you can't really
00:23:19.940 | tell right away.
00:23:20.940 | Yeah, the thing is, even if you get up on stage and tell people how Sophia's working,
00:23:27.500 | and then they talk to her, they still attribute more agency and consciousness to her than
00:23:34.540 | is really there.
00:23:36.820 | So I think there's a couple levels of ethical issue there.
00:23:41.900 | One issue is, should you be transparent about how Sophia is working?
00:23:49.620 | And I think you should.
00:23:50.940 | And I think we have been.
00:23:54.220 | I mean, there's articles online, there's some TV special that goes through me explaining
00:24:01.120 | the three subsystems behind Sophia.
00:24:03.320 | You know, the way Sophia works is out there much more clearly than how Facebook's AI works
00:24:10.540 | or something, right?
00:24:11.540 | I mean, we've been fairly explicit about it.
00:24:13.820 | The other is, given that telling people how it works doesn't cause them to not attribute
00:24:20.140 | too much intelligence agency to it anyway, then should you keep fooling them when they
00:24:26.980 | want to be fooled?
00:24:28.980 | And I mean, the whole media industry is based on fooling people the way they want to be
00:24:34.380 | fooled.
00:24:35.380 | And we are fooling people 100% toward a good end.
00:24:41.300 | I mean, we are playing on people's sense of empathy and compassion so that we can give
00:24:48.140 | them a good user experience with helpful robots, and so that we can fill the AI's mind with
00:24:55.820 | love and compassion.
00:24:57.300 | So I've been talking a lot with Hanson Robotics lately about collaborations in the area of
00:25:04.220 | medical robotics.
00:25:05.220 | And we haven't quite pulled the trigger on a project in that domain yet, but we may well
00:25:11.100 | do so quite soon.
00:25:12.580 | So we've been talking a lot about, you know, robots can help with elder care, robots can
00:25:18.320 | help with kids.
00:25:19.320 | David's done a lot of things with autism therapy and robots before.
00:25:24.400 | In the COVID era, having a robot that can be a nursing assistant in various senses can
00:25:28.480 | be quite valuable.
00:25:30.160 | The robots don't spread infection, and they can also deliver more attention than human
00:25:34.500 | nurses can give, right?
00:25:35.760 | So if you have a robot that's helping a patient with COVID, if that patient attributes more
00:25:42.740 | understanding and compassion and agency to that robot than it really has, because it
00:25:47.180 | looks like a human, I mean, is that really bad?
00:25:50.700 | I mean, we can tell them it doesn't fully understand you, and they don't care because
00:25:54.800 | they're lying there with a fever and they're sick.
00:25:57.200 | But they'll react better to that robot with its loving, warm facial expression than they
00:26:01.700 | would to a pepper robot or a metallic-looking robot.
00:26:05.960 | So it's really, it's about how you use it, right?
00:26:09.160 | If you made a human-looking, like, door-to-door sales robot that used its human-looking appearance
00:26:14.800 | to scam people out of their money, then you're using that connection in a bad way, but you
00:26:22.120 | could also use it in a good way.
00:26:24.980 | But then that's the same problem with every technology, right?
00:26:29.560 | Beautifully put.
00:26:30.540 | [laughs]
00:26:31.040 | [end]
00:26:33.040 | [music]
00:26:35.040 | [music ends]
00:26:37.040 | [no audio]
00:26:39.040 | [no audio]
00:26:41.040 | [no audio]
00:26:43.040 | [no audio]
00:26:45.040 | [no audio]
00:26:47.040 | [BLANK_AUDIO]