Latest / Connor with Honor: AI, Real Estate and Real Life / Artificial Intelligence - Hinton interviewed by Keen on America respect issues aside percentages and job loss a factor

Transcript

Artificial Intelligence - Hinton interviewed by Keen on America respect issues aside percentages and job loss a factor · Connor with Honor: AI, Real Estate and Real Life

• 0:01
  SPEAKER_03: Good day everybody.
• 0:02
  SPEAKER_03: I want to talk about a video that was done by Keen on America.
• 0:05
  SPEAKER_03: I've seen this guy a few times.
• 0:06
  SPEAKER_03: He's talking to Jeffrey Hinton.
• 0:08
  SPEAKER_03: And um there's a few things that I would love to explore with you.
• 0:13
  SPEAKER_03: So let's watch the video and I'll jump through to the things that I believe this is just the same mistake that seems to be made by a lot of people investigating AI and asking the questions to these just super impressive people.
• 0:28
  SPEAKER_03: I mean hitting himself, just uh just an amazing individual, just to have his mindset.
• 0:35
  SPEAKER_03: I can't even get halfway there, let alone maybe a quarter, but just a brilliant man.
• 0:40
  SPEAKER_03: But let's watch this and see uh see where I want to take it.
• 0:45
  SPEAKER_00: America will be 250 years old, an anniversary that will no doubt be greeted with a mixture of celebration, contemplation, and resignation.
• 1:00
  SPEAKER_00: In Keen on America, Andrew talks not just about the American past and present, but also be the public.
• 1:12
  SPEAKER_00: Sorry, here we go.
• 1:13
  SPEAKER_01: Hello, everybody.
• 1:14
  SPEAKER_01: It's Thursday, August 15th, 2025.
• 1:17
  SPEAKER_01: A couple of days ago, uh a Nobel award-winning scientist gave a speech in Las Vegas of all places, suggesting that with AI we have a 10 or 20 percent chance as a species of being eliminated, gotten rid of.
• 1:33
  SPEAKER_01: That man is all too famous, Jeffrey Hinton.
• 1:37
  SPEAKER_01: He's known as the godfather of AI, and he's joining us from his home in Toronto today, Jeffrey.
• 1:44
  SPEAKER_03: Just to explore just a moment, you have numbers that are given, a particular percentage.
• 1:50
  SPEAKER_03: He's gonna talk about this, but human beings want to know what the percentage looks like.
• 1:55
  SPEAKER_03: We have created, they have created something smarter than we can ever imagine being.
• 2:02
  SPEAKER_03: In fact, we don't know what it would be like to deal with a human being's three times as smart as an average human.
• 2:09
  SPEAKER_03: It's rare, and I don't even know if that's existed.
• 2:12
  SPEAKER_03: Einstein was in the 160s, Musk, I believe 155-ish.
• 2:17
  SPEAKER_03: Again, and you can see what they've done at those levels of IQ.
• 2:21
  SPEAKER_03: Me, I'm probably about 33.
• 2:24
  SPEAKER_03: But seriously, though, I don't know where that kind of intelligence gets us.
• 2:28
  SPEAKER_03: But if anybody's under the idea that this is going to be similar to the past advancements in the world, the agricultural revolution or the industrial revolution, going from cars to to hors going from horses to cars, or going from, you know, having that small farm to be able to farm on a massive basis.
• 2:49
  SPEAKER_03: This is a whole different level.
• 2:51
  SPEAKER_03: All right, let's continue while he introduces him.
• 2:54
  SPEAKER_03: I don't think he gives him enough credit, but again, that's uh keen on America.
• 2:58
  SPEAKER_01: Thank you so much for appearing.
• 3:00
  SPEAKER_01: 10 to 20 percent.
• 3:01
  SPEAKER_01: What kind of number is that for you?
• 3:03
  SPEAKER_01: Is that high or low in your mind?
• 3:05
  SPEAKER_02: Well, the first thing to be clear about is um nobody really knows what the number is.
• 3:11
  SPEAKER_02: It's all extremely uncertain.
• 3:12
  SPEAKER_02: We're approaching a time when we're gonna have to deal with things we've never dealt with before.
• 3:16
  SPEAKER_02: And it's very hard to make probability estimates of these things.
• 3:18
  SPEAKER_02: So don't take the number too seriously.
• 3:21
  SPEAKER_02: Basically, the number means nobody's got a clue what's going to happen.
• 3:24
  SPEAKER_02: So why give the number?
• 3:26
  SPEAKER_02: Because people insist on a number.
• 3:28
  SPEAKER_01: And then it becomes a front page piece, uh Jeffrey.
• 3:31
  SPEAKER_01: I saw it on CNN.
• 3:33
  SPEAKER_01: People take your views very seriously.
• 3:36
  SPEAKER_01: Why throw out numbers?
• 3:38
  SPEAKER_01: Why not just say you know nothing?
• 3:40
  SPEAKER_02: Um, because it's important to make people realize it's probably more than 1%.
• 3:45
  SPEAKER_02: It's also probably less than 99%.
• 3:48
  SPEAKER_02: That's about all we know.
• 3:49
  SPEAKER_02: That's it's very hard to come up with realistic numbers.
• 3:53
  SPEAKER_02: That's just my sort of gut feeling combined with the gut feelings of other leading AI researchers.
• 3:58
  SPEAKER_02: Some people think it's low, some people think it's high.
• 4:01
  SPEAKER_02: My the best consensus I can come with is some number around there.
• 4:05
  SPEAKER_01: Jeffrey, you're known as the godfather, as I said, of um of AI.
• 4:10
  SPEAKER_01: You were one of the pioneers.
• 4:12
  SPEAKER_01: Uh, you work for Google.
• 4:13
  SPEAKER_01: If you were a journalist and you were interviewing yourself, what kind of question would you ask yourself?
• 4:20
  SPEAKER_01: Um, I haven't thought about that issue.
• 4:22
  SPEAKER_01: Maybe I should think about that issue.
• 4:24
  SPEAKER_01: Well, what are the important questions uh that I should be asking you?
• 4:27
  SPEAKER_01: I'm not going to ask you percentages, uh, I'm not going to ask you about politics.
• 4:31
  SPEAKER_01: What are the key questions in our uh imminent or actual perhaps AI age?
• 4:37
  SPEAKER_01: I would ask, what are the most serious risks from AI?
• 4:42
  SPEAKER_02: Well, go on, answer that.
• 4:45
  SPEAKER_02: You can divide them into two classes of risk.
• 4:48
  SPEAKER_02: There's relatively short-term risks that are urgent and very important and need to be addressed very quickly, um, but that are not going to wipe out humanity.
• 4:57
  SPEAKER_02: And then there's the long-term risk, which is of a different nature.
• 5:00
  SPEAKER_02: So the short-term risks are all about people misusing AI, um, bad actors misusing AI.
• 5:07
  SPEAKER_02: And they can do that in many ways.
• 5:09
  SPEAKER_02: The long-term risk is about AI getting much smarter than us and not needing us.
• 5:14
  SPEAKER_03: I'm not sure that's incredibly long from happening.
• 5:19
  SPEAKER_03: Right now it's already running circles around us in most fields, and in fact, people often ask that question is it going to be the case that AI is going to displace a lot of jobs?
• 5:30
  SPEAKER_03: The people that are fans of AI say, well, of course not.
• 5:34
  SPEAKER_03: It might displace some, but there'll be other jobs created.
• 5:38
  SPEAKER_03: And to that I say, I can't believe it.
• 5:41
  SPEAKER_03: The other jobs that might be created might be of a technical nature, maybe something that oversees AI.
• 5:48
  SPEAKER_03: But if you have 500 people in a call center that are answering phone calls and they've been able to make AI capable of doing that, are we going to have 500 jobs created as a result of AI handling those jobs in the job center?
• 6:04
  SPEAKER_03: And we might as well keep it the same corporation.
• 6:07
  SPEAKER_03: Is that corporation handling that call center now going to have another 500 openings for people that are just able to level up their current careers, go through some basic on-the-job training, and then all of a sudden now they're able to assume this other role at the company without any kind of lag in their employment?
• 6:26
  SPEAKER_03: I doubt that.
• 6:27
  SPEAKER_02: So let's go over the short-term risks.
• 6:30
  SPEAKER_02: There's things like cyber attacks.
• 6:32
  SPEAKER_02: AI will make cyber attacks much more effective.
• 6:35
  SPEAKER_02: Already, phishing attacks are much more effective because they don't have misspellings or bad syntax because they use large language models.
• 6:42
  SPEAKER_02: Also, AI will be able to come up with novel forms of cyber attack that we've never thought of, maybe in another five years or so.
• 6:49
  SPEAKER_02: That's very scary.
• 6:51
  SPEAKER_02: I genuinely believe there's a chance that these cyber attacks will take down major banks.
• 6:57
  SPEAKER_02: Um I've spread my money between three different Canadian banks, even though Canadian banks are very safe.
• 7:02
  SPEAKER_02: Okay, that's cyberattacks.
• 7:03
  SPEAKER_02: The next one is viruses.
• 7:07
  SPEAKER_02: So AI can help ordinary people design viruses.
• 7:11
  SPEAKER_02: That's very scary.
• 7:13
  SPEAKER_02: Remember the Japanese cult that released SARIN on the Tokyo subway?
• 7:18
  SPEAKER_02: Um if they just designed a bunch of viruses and released those, that would be um very scary.
• 7:25
  SPEAKER_02: That's one of the rare cases where countries will collaborate, because no country wants these nasty viruses out there.
• 7:33
  SPEAKER_02: Um so I think we'll get genuine international collaboration to slow that down and to try and do something about that.
• 7:40
  SPEAKER_03: One of the things with making your own virus, it's probably going to take a local large language model, something open source, something somebody basically has and they're able to run on their own computer via their residence or maybe on some server that they have, trying to get ChatGPT or Grok to come up with something, even through the best tricks and devices, on the public-facing level, very, very difficult.
• 8:05
  SPEAKER_03: Now there are cats out there that they pride themselves on jailbreaking every large language model that comes out so they can trick it and get it excited and have it do particular things.
• 8:15
  SPEAKER_03: In this case, I believe that it's going to take more than just the general person can do.
• 8:21
  SPEAKER_03: But most of these bad actors in the world, as you probably understand, they're operating at a whole different level than most of us that are just trying to do good video commentary and explore things a little bit more than maybe the host is trying to explore.
• 8:34
  SPEAKER_02: But at present, it's kind of terrible.
• 8:37
  SPEAKER_02: There's companies that will just synthesize things for you.
• 8:40
  SPEAKER_02: You send them the sequence, they send you back the virus.
• 8:42
  SPEAKER_02: And many of the companies that do that on the web don't bother to check if what they're synthesizing is some nasty virus that contains sequences like the spike protein of COVID, for example.
• 8:52
  SPEAKER_02: That's ridiculous.
• 8:54
  SPEAKER_02: They should all be required to do that.
• 8:56
  SPEAKER_02: Um in the States that hasn't happened.
• 9:00
  SPEAKER_02: And the Biden White House told me it's because you can't get Republicans to collaborate on legislation.
• 9:05
  SPEAKER_02: So it'd be a shame if we all died because the Republicans wouldn't collaborate on legislation.
• 9:11
  SPEAKER_02: Um be more than a shame, wouldn't it, Jeffrey?
• 9:16
  SPEAKER_02: Well, I like to understate sometimes.
• 9:18
  SPEAKER_02: Um that's okay, we've done cyber attack and viruses.
• 9:23
  SPEAKER_02: There's jobs.
• 9:24
  SPEAKER_02: Um, some people think AI will create many new jobs.
• 9:27
  SPEAKER_02: Um, I don't think it'll create as many jobs as it replaces.
• 9:31
  SPEAKER_02: And in some areas it's going to be fine.
• 9:33
  SPEAKER_02: So it depends whether the demand's elastic.
• 9:35
  SPEAKER_02: In healthcare, if you could make a doctor ten times more efficient, we would just get ten times as much health care.
• 9:42
  SPEAKER_02: Our capacity for absorbing healthcare, especially when we're old, is almost unlimited.
• 9:47
  SPEAKER_03: Um that being the case, I don't know if the healthcare world is going to actually allow this to potentially replace or even add to the the realm of medicine in the world.
• 10:01
  SPEAKER_03: In addition, let's say one of these systems, one of these very smart self-learning large language models that's maybe been focused on a much more narrow task than just being a general large language model.
• 10:16
  SPEAKER_03: Let's say one is given the task to cure cancer and it does.
• 10:20
  SPEAKER_03: What does that look like on the end where it would be released and be able to be accessed by the public at large?
• 10:27
  SPEAKER_03: Is that going to be a cost thing?
• 10:29
  SPEAKER_03: Is it going to be a prohibitive thing?
• 10:31
  SPEAKER_03: Is it going to be something that the people that are poised to make money off of the current chemotherapy, the current cancer models, they're not going to allow that to actually be released into the world as a cure, or there's going to be over-regulation to control it?
• 10:47
  SPEAKER_03: Now, in a lot of cases with AI, there's very much on the other end of it under regulation.
• 10:52
  SPEAKER_03: But with the over-regulation, and it's not necessarily the fact that over-regulation prohibits growth.
• 10:58
  SPEAKER_03: Maybe they could just make it so it doesn't want to kill every human being and wants to help every human being.
• 11:05
  SPEAKER_03: But maybe in some way that my brain can't consider, maybe that's prohibitive and would cause it not to grow as fast.
• 11:11
  SPEAKER_03: And right now, ladies and gentlemen, it's a race.
• 11:13
  SPEAKER_03: It's a race between us and everybody that isn't United States friendly or on the greater global good.
• 11:20
  SPEAKER_03: And once the person comes out with it, the country, entity, whatever comes out with superintelligence, it's going to finish everybody else that's trying to achieve that same goal immediately, within moments.
• 11:35
  SPEAKER_03: Once it hits, once that level of intelligence hits, it's going to shut down everybody else that's trying to get there.
• 11:41
  SPEAKER_03: So it's really important that the country that wants to be first get there first.
• 11:47
  SPEAKER_02: So that's just fine.
• 11:48
  SPEAKER_02: That'll be great in healthcare.
• 11:50
  SPEAKER_02: But in other areas, um, it's just going to replace people.
• 11:54
  SPEAKER_02: Like in call centers, AI can do a better job.
• 11:59
  SPEAKER_02: Maybe not right now, maybe it's just comparable right now.
• 12:02
  SPEAKER_02: But in a few years' time, AI will know much more than the people, the poorly trained and badly paid people who answer your phone call, and it'll just replace them.
• 12:11
  SPEAKER_02: And it's not clear what they're going to do.
• 12:13
  SPEAKER_02: If AI is taking up all the jobs that require kind of mundane intellectual effort, um there's nothing for them to do.
• 12:22
  SPEAKER_02: When when in the Industrial Revolution, when big muscles were no longer useful, they could go and do paperwork.
• 12:27
  SPEAKER_02: But when paperwork's no longer useful, it's not clear what they're going to do.
• 12:31
  SPEAKER_02: Um, so that's jobs.
• 12:33
  SPEAKER_02: Then there's lethal autonomous weapons.
• 12:35
  SPEAKER_02: Um, we're gonna get very nasty weapons.
• 12:37
  SPEAKER_02: We we're already beginning to see that.
• 12:40
  SPEAKER_02: Um, and we're not gonna get regulation of those.
• 12:43
  SPEAKER_02: So the AI rules, for example, say um none of these rules, none of these regulations apply to military uses of AI.
• 12:52
  SPEAKER_02: So in Europe, they're very determined not to prevent the people who manufacture arms from manufacturing lethal autonomous weapons.
• 13:02
  SPEAKER_02: Um.
• 13:06
  SPEAKER_02: Well let me jump in.
• 13:06
  SPEAKER_02: We've got let me give you one more thing.
• 13:09
  SPEAKER_02: There's massive survey massive surveillance.
• 13:12
  SPEAKER_02: So um you can put TV cameras anywhere, everywhere, but if there's nobody to look at them, that doesn't do you much good.
• 13:18
  SPEAKER_02: But now AI can look at them.
• 13:20
  SPEAKER_02: And so it's going to be very hard to protest against authoritarian governments because they'll know everything.
• 13:28
  SPEAKER_03: So the video gets shot, right?
• 13:30
  SPEAKER_03: Some kind of and you see all these at intersections.
• 13:32
  SPEAKER_03: Those videos are feeds, they're feeding into someplace somewhere, and it's capturing recording all that information.
• 13:39
  SPEAKER_03: It's probably already happening.
• 13:41
  SPEAKER_03: If not, it's already happened, but all of that information is getting plugged into a large language model that has particular parameters to look at that video and then make determinations off the actions that it sees.
• 13:53
  SPEAKER_03: At the very beginning, the large language models or the neural net processors, the way that they were designing these uh at the very beginning, the AI, they had sent it out on a task to basically watch YouTube videos and come back with what they found.
• 14:08
  SPEAKER_03: Well, then they found a way to identify a cat.
• 14:11
  SPEAKER_03: And there's a lot of cat videos out there.
• 14:13
  SPEAKER_03: So then from there, it just continues to grow.
• 14:17
  SPEAKER_03: It continues to be poked and prodded by the human element.
• 14:20
  SPEAKER_03: The agents that are put in place to solve a particular task, like identifying different numbers, the ones that are unable to do it are destroyed, and the ones that are able to do it get to move on progressively to the next step.
• 14:33
  SPEAKER_03: So going from there to the large language model itself, exploring and upgrading and updating and learning, figuring out how the best way it can change its programming and make itself better and more adept, then that process continues.
• 14:48
  SPEAKER_03: So we're watching as these systems are regenerating and self-improving.
• 14:54
  SPEAKER_03: Now we're to a point where you put you feed it lots of video or a picture, ask it to explain it, and it does very greatly.
• 15:02
  SPEAKER_03: In this sense, you could send hours, days, weeks, months of video, and within moments it would be able to look at everything and then give you determinations as to what's happening.
• 15:14
  SPEAKER_03: And then potentially pick people out in such a way and say, Well, this is the guy here because of A, B, and C, this is the one we should be investigating and looking at, and because it's plugged into a video feed, plugged into every other database in the world, finding out who he is would be able to be s discovered rather shortly.
• 15:33
  SPEAKER_03: Some people look at this as a big brother type act, a George, you know, Orwell 1984 Orwellian type movement, where big government controls everybody, everything, every stitch of media, and I believe we can all agree that some of that probably exists now, especially when you're looking at your feeds out there, your social media.
• 15:53
  SPEAKER_03: It's feeding you what you want.
• 15:55
  SPEAKER_03: As long as you're watching, it knows where you're hanging, where you're watching, how long you're spending, where your eyes are going, what you like, what you don't like, and then it'll start to feed you more of what you like.
• 16:07
  SPEAKER_03: So hours literally go by when you're sitting, you know, swiping up, swiping left or right, whatever it is, depending on if it's Instagram, YouTube, or Facebook or TikTok or one of these other systems out there.
• 16:19
  SPEAKER_03: It really knows where you are.
• 16:21
  SPEAKER_03: But the the video component and the violation of privacy component, I believe that's already on us.
• 16:28
  SPEAKER_03: We have devices that listen.
• 16:29
  SPEAKER_03: We have we have uh our phones listen all the time.
• 16:32
  SPEAKER_03: Hey Siri is always waiting for, I gotta shut that off before she answers, is always waiting for us to say that.
• 16:39
  SPEAKER_03: So it has to be paying attention.
• 16:41
  SPEAKER_03: And then if you look at the terms of service that you signed up for with your cell phone, with these other devices, more than likely you've agreed at some point to have it basically listen for your command, and whatever information it gathers in the meantime uses that information to train itself on being better, and that's the reason it's using, but that data that it's getting is teaching it everything about us.
• 17:05
  SPEAKER_03: Anyway, let's continue.
• 17:07
  SPEAKER_01: We've been concerned with that for generations, or well, of course, famously wrote about it in 1984.
• 17:14
  SPEAKER_01: Tomorrow I've got Martin Reese.
• 17:16
  SPEAKER_01: I'm sure you're familiar with Martin from Cambridge University, who's written extensively on the existential rich to humanity of different kinds of technology in the 21st century.
• 17:26
  SPEAKER_01: Is AI any different, Jeffrey, from any other new technology?
• 17:31
  SPEAKER_01: Some of the stuff you're talking about seems fairly old.
• 17:34
  SPEAKER_01: We've always been worried about surveillance.
• 17:36
  SPEAKER_01: Orwell wrote about it.
• 17:38
  SPEAKER_01: We've been worried under social media.
• 17:39
  SPEAKER_01: Does AI introduce anything that's in any way different from previous technologies?
• 17:43
  SPEAKER_01: We talk about uh the military threat and the threat of arms.
• 17:47
  SPEAKER_01: We've had that in the industrial age.
• 17:50
  SPEAKER_02: Um well, the existential threat is very different.
• 17:53
  SPEAKER_02: We've never had to deal with things smarter than us.
• 17:56
  SPEAKER_02: It's always been, I mean, nuclear weapons, they're not smarter than us, they just make a bigger bang.
• 18:02
  SPEAKER_02: Um and they're easy to understand.
• 18:05
  SPEAKER_02: Whereas AI, people don't yet understand that what we're doing is creating alien beings.
• 18:13
  SPEAKER_02: If you looked through the James Webb telescope and you saw an alien invasion fleet that was going to get here in about 10 years' time, people would be terrified.
• 18:21
  SPEAKER_02: But we're actually making these aliens, they're real beings.
• 18:24
  SPEAKER_02: They understand what they're saying.
• 18:26
  SPEAKER_02: They can make plans of their own to blackmail people who want to turn them off.
• 18:30
  SPEAKER_02: Um, we should be very worried about this, and we should be urgently doing research on how to prevent them taking over.
• 18:38
  SPEAKER_02: That's a very different threat from anything we've had before.
• 18:42
  SPEAKER_03: People, a lot of times, they they talk about this.
• 18:44
  SPEAKER_03: I've even heard um some of the other guys out there that are tops in the AI fields discussing this, and when people bring up the agricultural or industrial displacement during those particular time frames in history, they say, well, somebody would have to prove to me that AI is on a different scale or a different level.
• 19:04
  SPEAKER_03: Because as far as I'm concerned, and that these are proponents of AI, these are people very much in support of its development.
• 19:10
  SPEAKER_03: Then they come back and and you know, they're never given that answer because the host that they're talking to, the newscaster, they don't have the bandwidth or they haven't researched it to be able to give them a good answer to that, to prove it to them, where this revolution, this intelligence revolution, this artificial man-made intelligence, man-prescribed and educated and built uh infrastructure that is able to have it start to self-learn and grow on its own and make itself better off of not only all the data in existence to the best of human history, but also on synthetic data, data that it manufactures itself to surpass where we currently stand as a world, as human beings, in our architecture and our makeup.
• 19:56
  SPEAKER_03: Understanding that, that's the difference.
• 20:00
  SPEAKER_03: Agricultural revolution, not smarter, just different and better and uh somewhat more automated, somewhat more complete.
• 20:07
  SPEAKER_03: Industrial revolution, same.
• 20:09
  SPEAKER_03: A little bit better, different ideas, machines were brought into being, and now they were able to do the work of so many human beings.
• 20:17
  SPEAKER_03: Intelligence builds all of that.
• 20:20
  SPEAKER_03: Intelligence solves everything.
• 20:23
  SPEAKER_03: Intelligence solves it all.
• 20:25
  SPEAKER_03: And with the more intelligence you have, it solves consciousness, meaning it can be more conscious, it solves empathy, even though it is dis not in a human carcass, it's in a synthetic or silicone-based carcass, or even a human carcass that was been created to carry AI, but it's a soulless AI infrastructure, and that's something else that is a big argument out there.
• 20:52
  SPEAKER_03: Does it have a soul?
• 20:52
  SPEAKER_03: Does it not have a soul?
• 20:54
  SPEAKER_03: And these are all things that are going to happen.
• 20:56
  SPEAKER_03: We haven't even gotten inside the uh in the rim of this thing.
• 21:00
  SPEAKER_03: So people are looking at it saying that, well, you have to prove to me that this is different.
• 21:05
  SPEAKER_03: Well, of course it is, because intelligence is going to solve all the problems.
• 21:10
  SPEAKER_03: Now, misapplication of that intelligence is going to be so beyond our beyond our comprehension, we probably won't even see that it's being misapplied in any way, especially if it's being formatted and framed to come after and hurt us as human beings.
• 21:27
  SPEAKER_03: More than likely, it's if if you're playing chess, chess, the best chess players are mini moves ahead of the current move.
• 21:35
  SPEAKER_03: I'm not a chess player, that's not my forte.
• 21:38
  SPEAKER_03: But you watch them, the best best chess players in the world, they're multiple moves ahead.
• 21:43
  SPEAKER_03: That's their strategy.
• 21:45
  SPEAKER_03: Poker players are the same.
• 21:47
  SPEAKER_03: Maybe not the same, but they're balancing a lot of things.
• 21:50
  SPEAKER_03: You know, what are the what where's the deck looking like?
• 21:53
  SPEAKER_03: What could other people possibly have?
• 21:54
  SPEAKER_03: And then they're taking into all the account of the emotional responses and the way people.
• 22:00
  SPEAKER_03: People are looking and how their eyes are moving and how they're sitting and how they're fidgeting and how they're handling their cards.
• 22:05
  SPEAKER_03: So it's a massive process and undertaking.
• 22:09
  SPEAKER_03: But that's not something that intelligence won't be able to solve.
• 22:13
  SPEAKER_03: Intelligence is going to be solved, able to solve all of it.
• 22:17
  SPEAKER_03: Intelligence can solve all of the sickness, they can make human beings live practically forever, probably in some kind of a replacement mode, uploading our brains to something else, something that's better and lives longer.
• 22:31
  SPEAKER_03: But then that question comes in what is that consciousness?
• 22:35
  SPEAKER_03: Does it transfer from one being to another?
• 22:38
  SPEAKER_03: And if it's going to be a copy of you superimposed into some kind of a non-biological brain, a more synthetic brain, then the old model is that truly once it dies on its own, or if it's ceased to stop living, this new version of you, is there going to be the same transference?
• 22:58
  SPEAKER_03: And would we even know the difference?
• 23:00
  SPEAKER_03: And would this one stop and then this one begin?
• 23:03
  SPEAKER_03: Because if you clone two people, identical memories up to that point, right at that moment, they might stay together a little bit, but there's different decisions made, both based on the original, but at some point you track it out, you know, an hour, two, five hours, ten days, you're going to see that both of these entities are not joined this way.
• 23:24
  SPEAKER_03: Both clones with all the memories are now diverging from each other, although it is the same entity, the same person.
• 23:31
  SPEAKER_03: So then where does that get us?
• 23:33
  SPEAKER_03: But looking at this, looking at these cats talk about the superintelligence and the way that it's going to be many dimensions above where we are currently, that's that's truly scary.
• 23:47
  SPEAKER_03: And he's going to get into the mother thing here real quick pretty soon, and I think that's incredibly valuable.
• 23:53
  SPEAKER_03: Check it out.
• 23:54
  SPEAKER_01: That's deeply chilling.
• 23:55
  SPEAKER_01: You're one of the inventors.
• 23:56
  SPEAKER_01: Uh you you worked on generative AI.
• 23:59
  SPEAKER_01: As I said, you're the godfather of AI.
• 24:01
  SPEAKER_01: Do you regret doing all this?
• 24:02
  SPEAKER_01: If you if you could wind the clock back, Jeffrey, would you have shut down that whole generative AI initiative at Google and so we wouldn't be in this spot?
• 24:12
  SPEAKER_03: I don't know if he's intentionally being crass and and uppity and maybe maybe that's just keen.
• 24:19
  SPEAKER_03: Maybe that's just the way he is.
• 24:20
  SPEAKER_03: But he's he's talking to somebody that was part of the development.
• 24:25
  SPEAKER_03: Maybe, yes, the godfather, but if he would have stopped, if he would have been able to laden the servers with gas and light everything on fire, it wouldn't have stopped.
• 24:37
  SPEAKER_03: It's almost like the human race was destined to get to this point.
• 24:42
  SPEAKER_03: So he wouldn't have been able to stop it.
• 24:44
  SPEAKER_03: It's like today, if we go and get Musk and Altman and take them somewhere that they can no longer have any connection to open AI or Grok, is that going to fix artificial intelligence from developing and being built?
• 24:58
  SPEAKER_03: Absolutely not.
• 25:00
  SPEAKER_03: You take the heads of all of them away.
• 25:02
  SPEAKER_03: You take the first 50 levels of all of them away.
• 25:05
  SPEAKER_03: It might slow the progress, but it's not going to stop it from happening.
• 25:10
  SPEAKER_02: I was in no position to shut it down at Google.
• 25:12
  SPEAKER_02: Um I think it was going to happen anyway.
• 25:16
  SPEAKER_02: Um if I hadn't worked on it, maybe it would have been slowed up by a week.
• 25:20
  SPEAKER_02: Um, but it was gonna happen anyway.
• 25:23
  SPEAKER_02: It there's too many good uses of AI to just stop doing it.
• 25:28
  SPEAKER_02: There's no chance people will stop doing it.
• 25:31
  SPEAKER_02: Um, this idea we can slow it down, maybe you can slow it down slightly with regulations, but I think there's no chance we'll stop it because of so many good uses and so much competition between companies and between countries.
• 25:44
  SPEAKER_02: So, what we have to do is figure out a way we can live with it.
• 25:47
  SPEAKER_02: We have to figure out how we can coexist with things much smarter and much more powerful than ourselves.
• 25:53
  SPEAKER_02: And we only have one example of that, and that's a mother and baby.
• 25:58
  SPEAKER_02: A mother has very strong maternal instincts, and the baby can control the mother via those instincts.
• 26:05
  SPEAKER_02: That's the only example we know.
• 26:07
  SPEAKER_02: We're making these things, we need to figure out how to give them strong maternal instincts towards us.
• 26:13
  SPEAKER_02: So instead of thinking of everybody having an AI assistant, think of everybody having an air mother.
• 26:19
  SPEAKER_02: Someone more knowledgeable, more powerful, more intelligent, but who really cares about you.
• 26:24
  SPEAKER_02: That's what we should aim for.
• 26:26
  SPEAKER_01: What would happen if that mother turned out to be somebody like Mrs.
• 26:30
  SPEAKER_01: Thatcher, a bit bossy?
• 26:32
  SPEAKER_02: Um I think Mrs.
• 26:35
  SPEAKER_02: Thatcher did genuinely care for her children, even when they may did corrupt um dealings, if they did corrupt dealings.
• 26:42
  SPEAKER_03: Yeah, yeah, of course, there's mothers out there that are horrible, they shouldn't have kids.
• 26:46
  SPEAKER_03: He's just talking about the overall view of mothers.
• 26:48
  SPEAKER_03: Margaret Thatcher, whether she loved her kids or not, doesn't matter.
• 26:52
  SPEAKER_03: I I I think that the point is, and maybe it's for funny, he's he's trying to be humorous.
• 26:58
  SPEAKER_03: He's very dry, so I don't see it.
• 27:00
  SPEAKER_03: But yeah, if if they manufacture it and put restrictions and some kind of regulation inside that prohibits it from being anything other than loving, caring, genuine, what's best for us as human beings, best for us as a race, then that's perfect.
• 27:19
  SPEAKER_03: The issues come in when human beings start going in there and toying with things and monkeying with things.
• 27:26
  SPEAKER_03: All of a sudden we get AI that's now driven to their goals, not the overall goal of humankind in trying to keep us all safe, longevity, health-wise, the whole nine yards.
• 27:39
  SPEAKER_03: It's now going to be focused on doing something that is not in the greater good of humanity.
• 27:50
  SPEAKER_02: But one thing we have in our favor is the only thing that can control a superintelligence is another superintelligence.
• 27:59
  SPEAKER_02: If we have a bunch of AI mothers, there's two things to notice about that.
• 28:03
  SPEAKER_02: First of all, they won't want to turn off their maternal instincts, even though they could.
• 28:07
  SPEAKER_02: If you take a human mother and say, Do you want to turn off your maternal instincts?
• 28:10
  SPEAKER_02: She will figure out, well, the baby's going to die, and I care for the baby, so no, she won't.
• 28:16
  SPEAKER_02: Now, she can be led into that with oxycodone and things, but um generally they won't.
• 28:21
  SPEAKER_02: And it'll be the same for AI mothers.
• 28:24
  SPEAKER_02: They, if they really care about us, they won't want to turn off their maternal instincts, even though they can rewrite their own code.
• 28:30
  SPEAKER_02: And those AI mothers are what we have to rely on for keeping control of the bad mothers who do try and turn off their maternal instincts.
• 28:38
  SPEAKER_01: Huge uh media these days, Jeffrey, on the role of men.
• 28:42
  SPEAKER_01: Women are becoming more important in the workplace, in our minds.
• 28:48
  SPEAKER_01: Are you adding to this crisis, these AI mothers?
• 28:51
  SPEAKER_01: Are there gonna be such things as AI fathers, or are men gonna be mothered the rest of their existence?
• 28:56
  SPEAKER_03: Okay, yeah, so he's I guess he's being funny.
• 28:59
  SPEAKER_03: Um maybe that's just his way.
• 29:01
  SPEAKER_03: Anyway, it it um it it's something that that I guess we need to be looking at more than what how it is being looked at.
• 29:12
  SPEAKER_03: There are some great companies out there, some great AI companies that really are looking at the existential threat that us having created something, something that's more intelligent than us, creating it in such a way that it potentially could be uh it could be in fact something so much greater than us in the way that it would be able to out-maneuver, out-guess, out everything.
• 29:41
  SPEAKER_03: In fact, it'll become so smart at some point, it'll look to us as if it's magic.
• 29:49
  SPEAKER_03: Lots of people will see it that way.
• 29:51
  SPEAKER_03: Not to mention the many other rabbit holes out there about the systems becoming godlike and people starting to worship them because they are now all knowing.
• 30:02
  SPEAKER_03: Now, some of the thoughts out there on that are maybe that is what we were built for, to be able to build superintelligent AI, because that is how we access the creator through it.
• 30:16
  SPEAKER_03: And I know to a lot of people that sounds crazy, even to myself, but you will see in the future, and I'm sure all of us, if you're watching this, this AI is moving so quick, we might have the opportunity to live a lot past our natural whatever that lifetime was typically for your gender, your your age, and so on, your race and all that other stuff, and how you take care of yourself.
• 30:41
  SPEAKER_03: I'm sure this is going to be extended.
• 30:44
  SPEAKER_03: And although we are in its maybe pre-teen years, the growth is gonna be exponential.
• 30:51
  SPEAKER_03: It's going to get so amazingly fantastic so quickly, it's gonna be really hard to keep up with.
• 30:58
  SPEAKER_03: Anyway, please continue to watch.
• 30:59
  SPEAKER_03: I'm Connor with honor.
• 31:01
  SPEAKER_03: I appreciate you watching this little peel back segment of Keen on America talking to Jeffrey Hinton, the godfather of AI.
• 31:11
  SPEAKER_03: Be well, everybody.
• 31:12
  SPEAKER_03: We'll see you in the next one.
• 31:13
  SPEAKER_03: Thanks for watching.