In this episode of the podcast, I sit down with Ben Bergman to talk about artificial intelligence, cryptos and artificial intelligence in general. Ben is a professor at the Santa Fe Institute, an AI researcher, an author, a speaker, and an advocate of the complex system theory of intelligence. We talk about how to understand the nature of intelligence and how it relates to our understanding of the world, and what it means for us to be a part of it. We also talk about the benefits and dangers of artificial intelligence and what we can learn from it, and why we should be concerned about it. I hope you enjoy this episode, and that it makes you think about AI a little bit more carefully about what we think about it and how we should approach it in the context of our current understanding of it and the implications it has for us and for the world at large. If you like the show, please consider becoming a patron patron, and/or share it on your socials. I'll be looking out for the best spots to get your tickets to the next show! Thanks again for listening and supporting the show. -Ben Bergman Thank you so much for being kind enough to take the time to be on the podcast and to share it with the world! -Jon Sorrentino - Ben Bergmann Jon Bergman ( ) Timestamps: 1:00:00 - What does artificial intelligence mean to you? 3:30 - What is artificial intelligence? 4:15 - Is it real? 5: What does it mean? 6: Is it a new life form? 7:40 - What can we learn from AI? 8:00 9:00 | What is it do we know about it? 11:30 | What are we trying to do? 12:30 13:40 | Is it possible? 15:40 16:00 What is AI a new thing? 17:10 | How does it matter? 18:00 Is it going to be different from nature? 19:20 | How can we know what it is? 21:10 22: Does it matter to you know it s going to become more than a life form ? 23:30 Is it different from us? 26:20 27:30 Can it be more than just a thing we know?
00:00:28.000Well, I've been following you as well.
00:00:30.000I've been paying attention to a lot of your lectures and talks and different things you've done over the last couple days as well, getting ready for this.
00:00:38.000AI is either people are really excited about it or they're really terrified.
00:00:45.000Either people have this dismal view of these robots taking over the world, or they think it's going to be some amazing sort of symbiotic relationship that we have with these things that's going to evolve human beings past the monkey stage that we're at right now.
00:01:21.000For three decades, and I first started thinking about AI when I was a little kid in the late 60s and early 70s when I saw AIs and robots on the original Star Trek.
00:01:33.000So I guess I've had a lot of cycles to process the positives and negatives of it, whereas now suddenly most of the world is thinking through all this for the first time.
00:01:46.000And when you first wrap your brain around the idea that There may be creatures 10,000 or a million times smarter than human beings.
00:01:55.000At first, this is a bit of a shocker, right?
00:01:58.000And then, I mean, it takes a while to internalize this into your worldview.
00:02:02.000Well, there's also, I think, there's a problem with the term artificial intelligence, because it's intelligent.
00:02:33.000It's an imperfect word, but it's not going away.
00:02:37.000Well, my question is, like, are we married to this idea of intelligence and of life being biological, being carbon-based tissue and cells and blood or insects or mammals or fish?
00:02:53.000Do you think that it's entirely possible that What human beings are doing and what people that are at the tip of AI right now that are really pushing the technology, what they're doing is really creating a new life form.
00:03:10.000That just the same way we recognize wasps and buffaloes and artificial intelligence is just going to be a life form that emerges from the creativity and ingenuity of human beings.
00:03:55.000I mean, if we can create digital systems or quantum computers or femto computers or whatever it is manifesting the patterns of organization that constitute intelligence, I mean, then there you are.
00:04:18.000There may be more dimensions to it, but when you look at what constitutes intelligence, thinking, cognition, problem solving, you know, it's the pattern of organization, not the specific material as far as we can tell.
00:04:32.000So we can see no reason, based on all the science that we know so far, that That you couldn't make an intelligence system out of some other form of matter rather than the specific types of atoms and molecules that make up human beings.
00:04:47.000And it seems that we're well on the way to being able to do so.
00:04:52.000When you're studying intelligence, you're studying artificial intelligence, did you spend any time...
00:04:59.000Studying the patterns that insects seem to cooperatively behave with, like how leafcutter ants build these elaborate structures underground, and wasps build these giant colonies.
00:05:18.000I grew up with the philosophy of complex systems, which was championed by the Santa Fe Institute in the 1980s.
00:05:27.000And the whole concept that there's an interdisciplinary complex system science, which includes biology, cosmology, psychology, sociology, the sort of universal patterns of...
00:05:42.000And, you know, ants and ant colonies have long been a paradigm case for that.
00:05:48.000I used to play with the ant colonies in my backyard when I was a kid.
00:05:53.000And you'd lay down food in certain patterns.
00:05:55.000You'd see how the ants are laying down pheromones and the colonies are organizing it in a certain way.
00:06:01.000And that's an interesting self-organizing, complex system.
00:06:06.000On its own, it's lacking some types of adaptive intelligence that human minds and human societies have, but it has also interesting self-organizing patterns.
00:06:17.000This reminds me of the novel Solaris by Stanislaw Lem, which was published in the 60s, which was Really, quite a deep novel, much deeper than the movie that was made of it.
00:06:49.000But that didn't get all the deep points of the novel.
00:06:53.000The original novel, in essence, there's this...
00:06:56.000There's this ocean coating the surface of some alien planet, which has amazingly complex fractal patterns of organization, and it's also interactive, like the patterns of organization on the ocean respond based on what you do,
00:07:12.000and when people get near the ocean, it causes them to hallucinate things, and even causes them to see...
00:08:30.000And there could be many, many types of AIs...
00:08:34.000That we could build with many, many different properties.
00:08:37.000Some could be wonderful to human beings, some could be horrible to human beings, some could just be alien minds that we can't even relate to very well.
00:08:51.000So we have a very limited conception of what an intelligence is.
00:08:55.000If we just think by close analogy to human minds, and this is important if you're thinking about engineering or growing artificial life forms or artificial minds, because it's not just can we do this, it's what kind of mind are we going to engineer or evolve,
00:09:14.000and there's a huge spectrum of possibilities.
00:09:17.000Yeah, that's one of the reasons why I asked you that.
00:09:20.000If we had created, if human beings had created some sort of an insect, and this insect started organizing and developing these complex colonies like a leafcutter ant and building these structures underground, people would go crazy.
00:10:16.000We can manipulate things, poison the environment, we can blow up entire countries with bombs if we'd like to, and we can also do wild creative things like send signals through space and land on someone else's phone on the other side of the world almost instantaneously.
00:10:29.000We have incredible power, but we're also so limited by our biology.
00:10:36.000The thing I think people are afraid of, and I'm afraid of, but I don't know if it makes any sense, is that the next Welcome to my show!
00:11:08.000In order to advance our species that we're so connected to these things, but they're so...
00:11:31.000It's almost inevitable by this point that humanity is going to create synthetic intelligences with tremendously greater general intelligence and practical capability than human beings have.
00:11:48.000I mean, I think I know how to do that with the software I'm working on with my own team.
00:11:53.000If we fail, there's a load of other teams who I think are a bit behind us, but are going in the same direction now, right?
00:12:00.000So you guys feel like you're at the tip of the spear with this stuff?
00:12:02.000I do, but I also think that's not the most important thing from a human perspective.
00:12:08.000The most important thing is that humanity as a whole is quite close to this threshold event, right?
00:12:15.000How far do you think it's quite close?
00:12:17.000By my own gut feeling, 5 to 30 years, let's say.
00:12:22.000But if I'm wrong and it's a hundred years, like in the historical timescale, that sort of doesn't matter.
00:12:27.000It's like, did the Sumerians create civilization 10,000 or 10,050 years ago?
00:12:32.000Like, what difference does it make, right?
00:12:34.000So, I think we're quite close to creating superhuman, artificial, general intelligence.
00:12:43.000That's, in a way, almost inevitable, given where we are now.
00:12:48.000On the other hand, I think we still have some agency regarding whether this comes out in a way that respects human values and culture, which are important to us now, given who and what we are,
00:13:04.000or that Is essentially indifferent to human values and culture in the same way that we're mostly indifferent to chimpanzee values and culture at this point.
00:13:15.000And completely indifferent to insect values and culture.
00:13:18.000Not completely, if you think about it.
00:13:20.000I mean, if I'm building a new house...
00:13:23.000I will bulldoze a bunch of ants, but yet we get upset if we extinct an insect species, right?
00:13:28.000So we care to some level, but we would like the super AIs to care about us more than we care about insects or great apes.
00:14:14.000The AIs that we create, It may come and go, and that's the nature of the universe.
00:14:19.000But on the other hand, of course, in my heart, from my situated perspective as an individual human, like if some AI tried to annihilate my 10-month-old son, I would try to kill that AI, right?
00:14:36.000Situated in this specific species, place, and time, I care a lot about the condition of all of us humans, and so I would like to not only create a powerful general intelligence, but create one which is...
00:14:54.000Is going to be beneficial to humans and other life forms on the planet, even while in some ways going beyond everything that we are.
00:15:06.000And there can't be any guarantees about something like this.
00:15:10.000On the other hand, humanity has really never had any guarantees about anything anyway.
00:15:17.000Since we created civilization, we've been leaping into the unknown.
00:15:23.000One time after the other in a somewhat conscious and self-aware way about it from, you know, agriculture to language to math to the industrial revolution.
00:15:33.000We're leaping into the unknown all the time, which is part of why...
00:15:39.000We're where we are today instead of just another animal species, right?
00:15:44.000So we can't have a guarantee that AGI, artificial general intelligences we create, are going to do what we consider the right thing, given our current value systems.
00:15:56.000On the other hand, I suspect we can bias the odds in the favor of human values and culture, and that's something I've put a lot of thought and work into alongside the basic algorithms of artificial cognition.
00:16:16.000Is the issue that the initial creation would be subject to our programming, but that it could perhaps program something more efficient and design something?
00:16:26.000Like, if you build creativity into artificial general intelligence… I mean, you have to.
00:16:30.000I mean, generalization is about creativity, right?
00:16:35.000Yeah, but is the issue that it would choose to not accept our values, which it might find… Occurs with some continuity and respect for the previous one.
00:16:56.000So, I mean, I have four human kids now.
00:16:59.000One is a baby, but the other three are adults, right?
00:17:01.000And with each of them, I took the approach of trying to teach the kids what my values were, not just by preaching at them, but by entering with them into shared situations.
00:17:13.000But then, you know, when your kids grow up, They're going to go in their own different directions, right?
00:17:21.000They all have the same sort of biological needs, which is one of the reasons why we have these desires in the first place.
00:17:26.000Mostly, right, but yet there still is an analogy.
00:17:28.000I think the AIs that we create, you can think of as our mind children, and We're starting them off with our culture and values, if we do it properly, or at least with a certain subset of the whole diverse, self-contradictory mess of human culture and values.
00:17:46.000But you know they're going to evolve in a different direction, but you want that evolution to take place in a reflective and caring way, rather than a heedless way.
00:17:58.000Because if you think about it, The average human a thousand years ago, or even 50 years ago, would have thought you and me were like hopelessly immoral miscreants who would abandon all the valuable things in life, right?
00:18:25.000I mean, there's all these things that we take for granted now that Not that long ago, we're completely against what most humans considered maybe the most important values of life.
00:18:38.000So, I mean, human values itself is completely a moving target.
00:19:27.000My sister went to the high school prom with a black guy, and so we got our car turned upside down, the windows of our house smashed, and it was like a humongous thing, and it's almost unbelievable now, right?
00:19:39.000Because now, no one would care whatsoever.
00:19:59.000And what you want is for the evolution of the AI's values to be coupled closely with the evolution of human values, rather than Going off in some utterly different direction that we can't even understand.
00:20:14.000But this is literally playing God, right?
00:20:16.000I mean, if you're talking about, like, trying to program in values… I don't think you can program in values that fully.
00:20:24.000You can program in a system for learning and growing values.
00:20:29.000And here, again, the analogy with human kids is not hopeless.
00:20:34.000Like, telling Telling your kids, these are the 10 things that are important, doesn't work that well, right?
00:20:41.000What works better is you enter into shared situations with them, they see how you deal with the situations, you guide them in dealing with real situations, and that forms their system of values.
00:20:53.000And this is what needs to happen with AIs.
00:20:56.000They need to grow up entering into real-life situations.
00:21:00.000With human beings, so that the real-life patterns of human values, which are worth a lot more than the homilies that we enunciate formally, right?
00:21:10.000The real-life pattern of human values gets inculcated into the intellectual DNA of the AI systems.
00:21:17.000And this is part of what worries me about the way the AI field is going at this moment, because, I mean, most of the really powerful We're good to go.
00:21:51.000Right, if they don't have any problem morally and ethically with manipulating us, which we're very malleable, right?
00:22:35.000Yeah, you look in the US, it's like, somehow, you have laws that allow random lunatics to buy all the guns they want, and you have all these people getting shot.
00:22:46.000So, similarly, from the outside, you could look at it like, this species is creating the successor intelligence, and almost all the resources going into creating their successor intelligence are going into making AIs to do...
00:23:05.000Surveillance like military drones and advertising agents to brainwash people into buying crap they don't need.
00:23:31.000Well, yeah, the applications that are getting the most attention are the financial lowest hanging fruit, right?
00:23:37.000So, for example, among many projects I'm doing with my SingularityNet team, We're looking at applying AI to diagnose agricultural disease.
00:23:48.000So you can look at images of plant leaves, you can look at data from the soil and atmosphere, and you can project whether disease in a plant is likely to progress badly or not, which tells you, do you need medicine for the plant?
00:24:03.000This is an interesting area of application.
00:24:06.000It's probably quite financially lucrative in a way, but it's a more complex industry than selling stuff online.
00:24:25.000Yeah, yeah, but there's a lot of specific aspects, right?
00:24:29.000So, I mean, AI for medicine, again, there's been papers on machine learning applied to medicine since the 80s and 90s.
00:24:37.000But the amount of effort going into that compared to advertising or surveillance is very small.
00:24:44.000Now this has to do with the structure of the pharmaceutical business as compared to the structure of the tech business.
00:24:50.000So when you look into it, there's good reasons for everything, right?
00:24:56.000But nevertheless, the way things are coming down right now is certain...
00:25:04.000Biases to the development of early stage AIs are very marked, and you could see them.
00:25:11.000And I mean, I'm trying to do something about that together with my colleagues in SingularityNet, but of course, it's sort of a David versus Goliath thing.
00:25:22.000Well, of course you're trying to do something different, and I think it's awesome what you guys are doing.
00:25:27.000But it just makes sense to me that the first applications are going to be the ones that are more financially viable.
00:25:34.000Well, the first applications were military, right?
00:25:37.000I mean, until about 10 years ago, 85% of all funding into AI was from US plus Western Europe militaries.
00:25:45.000Well, what I'm getting at is that it seems that...
00:25:48.000Money and commerce are inexorably linked to innovation and technology because there's this sort of thing that we do as a culture where we're constantly trying to buy and purchase bigger and better things.
00:26:02.000We always want the newest iPhone, the greatest laptop, we don't want the coolest electric cars, whatever it is.
00:26:15.000Materialism, in a lot of ways, fuels innovation because this is It does, but I think there's an argument that as we approach a technological singularity, we need new systems.
00:26:27.000Because if you look at how things have happened during the last century, what's happened is that governments have funded most of the core innovation.
00:26:37.000I mean, this is well known that most of the technology inside a smartphone was funded by...
00:26:42.000U.S. government, a little about European government, GPS and the batteries and everything.
00:26:54.000And this process occurs with a certain time cycle to it, where government spends decades funding core innovation and universities, and then industry spends decades figuring out how to scale it up and make it palatable to users.
00:27:32.000So the gene is out of the bottle, essentially.
00:27:34.000Yeah, but we still need a lot of new, amazing, creative innovation to happen.
00:27:39.000But somehow or other, new structures are going to have to evolve to make it happen.
00:27:45.000And you can see everyone's struggling to figure out what these are.
00:27:48.000So this is why you have big companies embracing open source.
00:27:52.000Google releases TensorFlow, and there's a lot of...
00:27:56.000And I think some projects in the cryptocurrency world have been looking at that too.
00:28:01.000Like how do we use tokens to incentivize independent scientists and inventors to do new stuff without them having to be in a government research lab or in a big company.
00:28:13.000So I think we're going to need the evolution of...
00:28:17.000New systems of innovation and of technology transfer as things are developing faster and faster and faster.
00:28:26.000And this is another thing that's sort of gotten me interested in the whole decentralized world and the blockchain world is the promise of new modes of economic and social organization that can bring more of the world into the research process and accelerate the technology transfer process.
00:29:00.000It's big tech, which are advertising agencies in essence.
00:29:06.000Facebook, social media, things that are constantly predicting your next purchase, right?
00:29:10.000Yeah, because if you think about it, and I'm in...
00:29:15.000Even in a semi-democracy like we have in the US, I mean, those who control the brainwashing of the public, in essence, control who votes for what, and who controls the brainwashing of the public is advertising agencies,
00:29:31.000and who increasingly are the biggest advertising agencies are the big tech companies who are Accumulating everybody's data and using it to program their minds to buy things.
00:29:42.000So this is what's programming the global brain of the human race.
00:29:46.000And of course, there are close links between big tech and the military.
00:29:51.000Look, Amazon has, what, 25,000 person headquarters in Crystal City, Virginia, right next to the Pentagon.
00:29:57.000I mean, China, it's even more direct and unapologetic, right?
00:30:01.000So it's a new, like, military-industrial advertising complex, which is guiding the evolution of the global brain on the planet.
00:30:13.000We found that with this past election, right?
00:30:15.000With all the intrusion by foreign entities trying to influence the election, that these giant houses set up to write bad stories about whoever they don't want to be in office?
00:30:29.000Yeah, in a way, that's almost a red herring, but it, I mean, the Russian stuff is almost a red herring, but it revealed what the processes are, which are used to programming.
00:30:41.000Oh, because I think whatever programming of Americans' minds is done by the Russians is minuscule compared to the programming of Americans' minds by the Americans.
00:30:54.000American corporate and government elite, right?
00:30:56.000But it's fascinating that anybody's even jumping in as well as the American elite.
00:32:49.000First got me to spend a lot of time in China, but then I was doing some research at Hong Kong Polytechnic University, and then my good friend David Hansen was visiting me in Hong Kong.
00:33:01.000I introduced him to some investors there, which ended up with him bringing his company Hansen Robotics to Hong Kong.
00:33:08.000So now, after I moved there because of falling in love with Ray Ting, then I brought my friend David there, Then Hanson Robotics grew up there, and there's actually a good reason for Hanson Robotics to be there, because the best place in the world to manufacture complex electronics is in Shenzhen,
00:33:26.000right across the border from Hong Kong.
00:33:28.000So now I've been working there with Hanson Robotics on the Sophia robots and other robots for a while, and I've accumulated a whole AI team there around Hanson Robotics and SingularityNet.
00:33:40.000So I mean, by now I'm there because my whole AI and robotics teams are there.
00:34:53.000Going into too much detail, like when I was in D.C. working with various government agencies, it became clear there is tremendously more information obtained by government agencies than most people realize.
00:35:08.000This was true way before Snowden and WikiLeaks and all these revelations.
00:35:13.000And what is publicly understood now is...
00:35:19.000Probably not the full scope of the information that governments have either.
00:35:25.000So, I mean, privacy is pretty much dead.
00:35:28.000And David Brin, do you know David Brin?
00:35:42.000Years ago, called the Transparent Society, where he said there's two possibilities, surveillance and surveillance.
00:35:49.000It's like the power elite watching everyone, or everyone watching everyone.
00:35:54.000I think everyone watching everyone is inevitable.
00:35:56.000So he articulated this as essentially the only two viable possibilities, and he's like, we should be choosing and then creating which of these alternatives we want.
00:36:16.000I mean, it was well more than a decade ago.
00:36:18.000It's weird when some people just nail it on the head decades in advance.
00:36:22.000I mean, most of the things that are happening in the world now were foreseen by Stanislaw Lem, the Polish author I mentioned.
00:36:31.000Valentin Turchin, a friend of mine who was the founder of Russian AI, he wrote a book called The phenomenon of science in the late 60s.
00:36:38.000Then, you know, in 1971 or 2, when I was a little kid, I read a book called The Prometheus Project by a Princeton physicist called Gerald Feinberg.
00:36:49.000You read a physicist's book when you're five years old?
00:36:51.000Yeah, I started reading when I was two, and my grandfather was a physicist, so I was reading a lot of stuff then.
00:36:57.000But Feinberg, in this book, he said, you know, within the next few decades, humanity is going to create nanotechnology, it's going to create machines smarter than people, and it's going to create the technology to allow human biological immortality.
00:37:11.000And the question will be, do we want to use these technologies, you know, to promote rampant consumerism, or do we want...
00:37:18.000To use these technologies to promote, you know, spiritual growth of our consciousness into new dimensions of experience.
00:37:24.000And what Feinberg proposed in this book in the late 60s, which I read in the early 70s, he proposed the UN should send a task force out to go to everyone in the world, every little African village, and educate the world about nanotech, life extension,
00:37:40.000and AGI, and get the whole world to vote on whether we should develop these technologies toward consumerism, Or toward consciousness expansion.
00:37:58.000Then I tried to explain this to people and I'm like, oh shit, I guess it's going to be a while until the world catches on.
00:38:07.000So I instead decided I should build a spacecraft, go away from the world at rapid speed and come back after like a million years or something when the world was far more advanced.
00:38:20.000So now, well then you go away another million years and see what aliens have evolved.
00:38:24.000So now, Pretty much the world agrees that life extension, AGI, and nanotechnology are plausible things that may come about in the near future.
00:38:35.000The same question is there that Feinberg saw like 50 years ago, right?
00:38:42.000The same question is there, like, do we develop this for...
00:39:06.000On the other hand, there's the possibility that by bypassing governments in the UN and doing something decentralized, you can create a democratic framework, you know, within which, you know, a broad swath of the world can be involved in a participatory way in guiding the direction of these advances.
00:39:26.000Do you think that it's possible that instead of choosing, that we're just going to have multiple directions that it's growing in, that there's going to be consumer-based?
00:39:33.000There will be multiple directions, and that's inevitable.
00:39:38.000It's more a matter of whether… Anything besides the military advertising complex gets a shake, right?
00:39:46.000So I mean, if you look in the software development world, open source is an amazing thing, right?
00:40:41.000So, I mean, this hippie, crazy thing where no one owns the code...
00:40:46.000It didn't have to overtake the whole software economy and become everything to become highly valuable and inject a different dimension into things.
00:40:58.000And I think the same is true with decentralized AI, which we're looking at with singularity.
00:41:04.000We don't have to actually put Google and the US and Chinese military and Tencent Out of business, right?
00:41:13.000Although if that happens, that's fine.
00:41:15.000But it's enough that we become an extremely major player in that ecosystem so that this participatory and benefit-oriented aspect becomes a really significant component of how humanity is developing general intelligence.
00:41:36.000It's accepted, generally accepted, that human beings will consistently and constantly innovate.
00:41:41.000It just seems to be a characteristic that we have.
00:43:17.000And this is certainly one of the factors which is driving the creation of AI. I don't think that alone would make the creation of AI inevitable.
00:44:12.000My kids are Zarathustra, Amadeus, Zebulon, Ulysses, Scheherazade, and then the new one is CORXI, Q-O-R-X-I, which is an acronym for Quantum Organized Rational Expanding Intelligence.
00:44:45.000And to me, that's like the most important thing we could be applying AI to because, you know, mathematics is the key to all modern science and engineering.
00:45:31.000With our drive to create and innovate, and this becomes an almost unstoppable force within human society.
00:45:38.000And what we've seen in the last three to five years is suddenly national leaders and titans of industry, and even pop stars, right?
00:45:48.000They've woken up to the concept that Wow, smarter and smarter AI is real, and this is going to get better and better, like, within years to decades, not centuries to millennia.
00:46:01.000So, now the cat's out of the bag, nobody's going to put it back, and it's about, you know, how can we direct it in the most...
00:46:48.000On the other hand, I'd like to let one of me remain in human form, you know, get rid of death and disease and...
00:46:56.000Psychological issues and just live happily forever, you know, in the people's zoo, watched over by the machines of love and grace, right?
00:47:04.000So I mean, you can have, it doesn't have to be either or, because once you can scan your brain and body and 3D print new copies of yourself, you could have multiple of you explore different scenarios.
00:47:16.000Right, but isn't that a giant resource hog?
00:47:17.000There's a lot of mass energy in the universe.
00:47:34.000But wealthy people being able to reproduce themselves and just having this idea that they would like their ego to exist in multiple different forms, whether it's some super symbiote form, That's connected to artificial intelligence or some biological form that's immortal or some other form that stands just as a normal human being as we know it in 2018. Have you had multiple versions of yourself over and over and over again like that?
00:48:00.000Once you get to the point where you have a superhuman general intelligence that can do things like fully scan the human brain and body and 3D print more of them, by that point You're at a level where scarcity of material resources is not an issue at the human scale of doing things.
00:48:21.000Scarcity of human resources in terms of what the Earth can hold?
00:48:24.000Scarcity of mass energy, scarcity of molecules to print more copies of yourself.
00:48:29.000I think that's not going to be the issue at that point.
00:48:32.000But what people are worried about is environmental concerns of overpopulation.
00:48:35.000Because people are worried about what they see in front of their faces right now, but people are not...
00:48:41.000Are not thinking deeply enough about what potential would be there once you had superhuman AIs doing the manufacturing and the thinking.
00:48:56.000I mean, the amount of energy in a single grain of sand, if you had an AI able to appropriately leverage that energy is tremendously more than most people think.
00:49:09.000The amount of computing power in a grain of sand is like a quadrillion times all the people on Earth put together.
00:50:10.000There may be other problem issues that we can't even conceive at this moment, of course.
00:50:15.000But the intelligence would be so vastly superior to what we have currently that they'll be able to find solutions to virtually every single problem we have.
00:50:31.000People don't want to hear that, though.
00:50:33.000Environmental people don't want to hear that, right?
00:50:34.000Well, I mean, I'm also, on an everyday life basis, like, until we have these super AIs, I don't like the garbage washing up on the beach near my house either, right?
00:50:48.000On an everyday basis, of course, we want to promote health in our bodies and in our environments right now, as long as there's measurable uncertainty regarding when the benevolent super AIs will come about.
00:51:05.000The main question isn't whether once you have a beneficially disposed super AI, it could solve all our current petty little problems.
00:51:13.000The question is, can we wade through the muck of modern human society and psychology to create this beneficial super AI? Right.
00:51:25.000I believe I know how to create a beneficial super AI, but it's a lot of work to get there.
00:51:32.000And of course, there's many teams around the world working on vaguely similar projects now, and it's not obvious what kind of super AI... We're actually going to get once we get there.
00:51:46.000Yeah, it's all just guesses at this point, right?
00:51:48.000It's more or less educated guesses, depending on who's doing the guessing.
00:51:52.000Would you say that it's almost like we're in a race of the primitive primate biology versus the potentially beneficial and benevolent artificial intelligence that the best aspects of this primate can create?
00:52:09.000Is it the warmongers and the greedy whores that are smashing the world under its boots?
00:52:14.000Or is it the scientists that are going to figure out some super intelligent way to solve all of our problems?
00:52:20.000I look at it more as a struggle between different modes of social organization than individual people.
00:52:32.000When I worked in D.C. with intelligence agencies...
00:52:36.000Most of the people I met there were really nice human beings who believed they were doing the best for the world, even if some of the things they were doing like I thought were very much not for the best of the world, right?
00:52:51.000So, I mean, military mode of organization or large corporations as a mode of organization are...
00:53:00.000In my view, not generally going to lead to beneficial outcomes for the overall species and for the global brain.
00:53:08.000The scientific community, the open source community, I think, are better modes of organization.
00:53:14.000And, you know, the better aspects of the blockchain and crypto community have a better mode of organization.
00:53:20.000So I think if this sort of open, decentralized mode of organization can...
00:53:28.000Marshall more resources as opposed to this centralized authoritarian mode of organization, then I think things are going to come out for the better.
00:53:38.000And it's not so much about bad people versus good people.
00:53:41.000You can look at, like, the corporate mode of organization is almost a virus that's colonized a bunch of humanity and is sucking people into working according to this mode.
00:53:52.000And even if they're really good people, and the individual task they're working on isn't bad in itself, they're working within this mode that's leading their work to be used for ultimately a non-good end.
00:54:06.000Yeah, that is a fascinating thing about corporations, isn't it?
00:54:10.000The diffusion of responsibility and being a part of a gigantic group that you as an individual don't feel necessarily connected or responsible to the ultimate group.
00:54:18.000Well, even the CEO isn't fully responsible.
00:54:21.000Like, if the CEO does something that isn't in accordance with the higher goals of the organization, they're just replaced, right?
00:54:28.000So, I mean, there's no one person who's in charge.
00:54:43.000In this way, in some ways, the Asian countries are a little more intelligent than Western countries, and the Asian governments realize the power of corporations to mold society, and there's a bit more feedback between the government and corporations,
00:56:00.000Politically, they were poorer than two-thirds of sub-Saharan African nations in the late 60s.
00:56:06.000And it is through the government intentionally stimulating corporate development toward manufacturing and electronics that they grew up.
00:56:16.000Now, I'm not holding that up as a great paragon for the future or anything, but it does show that there's...
00:56:25.000There's many modes of organization of people and resources other than the ones that we take for granted in the US. I don't think Samsung and LG are the ideal for the future either, though.
00:56:37.000I mean, I'm much more interested in, you know...
00:56:58.000Is the way to be ongoingly disruptive.
00:57:02.000And open source is a good example of that.
00:57:04.000Like, when the open source movement started, they weren't thinking about machine learning.
00:57:09.000But you know, the fact that open source is out there and is then prevalent in the software world, That paved the way for AI to now be centered on open source algorithms.
00:57:20.000So right now, even though big companies and governments dominate the scalable rollout of AI, the invention of new AI algorithms is mostly done by people creating new code and putting it on GitHub or GitLab or other open source repositories.
00:57:36.000Now, open source is self-explanatory in its title.
00:58:10.000The idea of a distributed ledger, which is basically like a distributed Excel spreadsheet or database.
00:58:16.000It's just a store of information, which is not stored just in one place, but there's copies of it in a lot of different places.
00:58:23.000Every time my copy of it is updated, everyone else's copy of it has got to be updated.
00:58:29.000And then there's various bells and whistles, like sharding, where it can be broken in many pieces, and each piece is stored many places or something.
00:58:38.000That's a distributed ledger, and that's just distributed computing.
00:58:41.000Now, what makes it more interesting is when you layer decentralized control onto that.
00:58:47.000So imagine you have this distributed Excel spreadsheet or distributed database.
00:58:52.000There's copies of it stored in a thousand places.
00:58:55.000But to update it, you need like 500 of those thousand people who own the copies to vote, yeah, let's do that update.
00:59:02.000So then you have a distributed store of data, and you have like a democratic voting mechanism to determine when all those copies can get updated together, right?
00:59:13.000So then what you have is a data storage and update mechanism that's controlled...
01:00:03.000Is when I vote, I don't have to say, yeah, this is Ben Gertzel voting for this update to be accepted or not.
01:00:09.000It's just ID number 1357264. And then encryption is used to make sure that, you know, it's the same guy voting every time that it claims to be without needing, like, your passport number or something,
01:00:43.000Where the blockchain comes from is like a data structure where to store the data in this distributed database, it's stored in a chain of blocks where each block contains data.
01:01:11.000It's one of those terms we're stuck with even though it's not quite technically...
01:01:16.000Not quite technically accurate anymore.
01:01:19.000I don't know another buzzword for it, right?
01:01:23.000What it is, it's a distributed ledger with encryption and decentralized control.
01:01:28.000And blockchain is the buzzword that's come about for that.
01:01:32.000What got me interested in blockchain really is this decentralized control aspect.
01:01:37.000So my wife, who I've been with for 10 years now, she dug up recently something I'd forgotten, which is a webpage I'd made in 1995, like a long time ago, where I'd said, hey, I'm going to run for president on the decentralization platform,
01:02:02.000So the idea of decentralized control seemed very important to me back then, which is well before Bitcoin was invented, because I could see a global brain is evolving on the planet, involving humans, computers, communication devices,
01:02:18.000and we don't want this global brain to be controlled by a small elite.
01:02:22.000We want the global brain to be controlled in a decentralized way.
01:02:26.000So that's really the beauty of this concept.
01:02:32.000What got me interested in the practical technologies of blockchain was really when Ethereum came out and you had the notion of a smart contract.
01:05:35.000So what happened in the last couple of years is a bunch of people realized you could use this Ethereum programming framework to create A new cryptocurrency, like a new artificial money,
01:05:51.000and then you could try to get people to use your new artificial money for certain types of...
01:07:07.000With a simple programming language, all sorts of transactions between people, companies, whatever, all sorts of exchanges of information.
01:07:16.000So, I mean, it's about decentralized voting mechanisms.
01:07:20.000It's about AIs being able to send data and processing for each other and pay each other for their transactions.
01:07:28.000So, I mean, it's about automating supply chains and shipping and e-commerce.
01:07:35.000In essence, just like computers and the internet started with a certain small set of applications and then pervaded almost everything, right?
01:07:46.000It's the same way with blockchain technology.
01:07:48.000It started with digital money, but the core technology is going to pervade Almost everything, because there's almost no domain of human pursuit that couldn't use, like, security through cryptography, some sort of,
01:08:04.000you know, participatory decision-making, and then distributed storage of information, right?
01:08:09.000And these things are also valuable for AI, which is how I got into it in the first place.
01:08:13.000I mean, if you're making a very, very powerful AI that is going to, you Through the practical value it delivers, you will grow up to be more and more and more intelligent.
01:08:26.000I mean, this AI should be able to engage a large party of people and AIs in participatory decision-making.
01:08:33.000The AI should be able to store information, you know, in a widely distributed way.
01:08:38.000And the AI certainly should be able to use, you know, security and encryption to validate who are the parties involved in its operation.
01:08:46.000And I mean, these are the key things behind blockchain technology.
01:08:51.000The fact that blockchain began with artificial currencies, to me, is a detail of history, just like the fact that the internet began as like a nuclear early warning system, right?
01:09:01.000I mean, it did, it's good for that, but as it happens, it's also even better for a lot of other things.
01:09:08.000Yeah, the solution for the financial situation that we find ourselves in, it's one of the more interesting Things about cryptocurrencies that someone said, okay, look, obviously we all kind of agree that our financial institutions are very flawed.
01:09:24.000The system that we operate under is very fucked up.
01:10:28.000Kids today, in particular, the ones that have grown up with the internet as a constant force in their life, I think they're more likely to embrace something along those lines.
01:10:38.000There's no doubt that cryptographic formulations of money are going to become the standard.
01:10:48.000Do you think that's going to be the standard?
01:11:00.000I mean, a government could just say, we will create this cryptographic token, which counts as a dollar.
01:11:06.000I mean, most dollars are just electronic anyway, right?
01:11:09.000So what habitually happens is technologies that are invented to subvert the establishment are converted to a form where they...
01:11:20.000Help bolster the establishment instead.
01:11:23.000I mean, in financial services, this happens very rapidly.
01:11:28.000Like PayPal, Peter Thielen, those guys started PayPal thinking they were going to obsolete fiat currency and make an alternative to the currencies run by nation states.
01:11:38.000Instead, they were driven to make it a credit card processing front end, right?
01:11:45.000One thing that could happen with cryptocurrency is it just becomes a mechanism for governments and big companies and banks to do their things more efficiently.
01:11:56.000So what's interesting isn't so much the digital money aspect, although it is in some ways a great way to do digital money.
01:12:04.000What's interesting is with all the flexibility it gives you to script complex computing networks, In there is the possibility to script new forms of participatory democratic self-organizing networks.
01:12:21.000So blockchain, like the internet or computing, is a very flexible medium.
01:12:26.000You could use it to make tools of oppression, or you could use it to make tools of amazing growth and liberation.
01:12:35.000And obviously we know which one I'm more interested in.
01:12:57.000Well, the heaviest uses of blockchain now are probably inside large financial services companies, actually.
01:13:05.000So if you look at Ethereum, the project I mentioned, so Ethereum is run by an open source, an open foundation, Ethereum Foundation.
01:13:15.000Then there's a consulting company called ConsenSys.
01:13:19.000which is a totally separate organization that was founded by Joe Lubin who was one of the founders of Ethereum in the early days and ConsenSys has funded a bunch of the work within the Ethereum foundation and community but ConsenSys has done a lot of contracts just working with governments and big companies to customize code based on Ethereum to help with their internal operations so actually a lot of the practical value has been With stuff that
01:13:49.000isn't in the public eye that much, but it's like back-end inside of companies.
01:13:55.000In terms of practical customer-facing uses of cryptocurrency, I mean, the Tron blockchain, which is different than Ethereum, that has a bunch of games on it, for example, and some online gambling, for that matter.
01:14:39.000And this is one of the things we're aiming at with our SingularityNet project is to, you know, by putting...
01:14:47.000AI on the blockchain in a highly effective way.
01:14:51.000And then we're also, we have these two tiers.
01:14:55.000So we have the SingularityNet Foundation, which is creating this open source decentralized platform in which AIs can talk to other AIs and, you know, like ants in the colony grouped together to form smarter and smarter AI. Then we're spinning off A company called the Singularity Studio,
01:15:12.000which will use this decentralized platform to help big companies integrate AI into their operations.
01:15:19.000So with the Singularity Studio company, we want to get all these big companies using the AI tools in the SingularityNet platform, and then we want to drive massive usage of blockchain in the SingularityNet platform.
01:15:36.000If we're successful with what we're doing, this will be within a year from now or something by far the biggest Usage of blockchain outside of financial exchange is our use of blockchain within SingularityNet for AI,
01:15:52.000basically for customers to get the AI services that they need for their businesses and then for AIs to transact with other AIs, paying other AIs for doing services for them.
01:16:08.000It's like a society and economy of minds.
01:16:10.000It's not like one monolithic AI. It's a whole bunch of AIs carried by different people all over the world, which not only are in the marketplace providing services to customers, but each AI is asking questions of each other and then...
01:16:24.000Rating each other of how good they are sending data to each other and paying each other for their services.
01:16:29.000So this network of AIs can emerge in intelligence on the whole network level as well as there being intelligence in each component.
01:16:39.000And is it also fascinating to you that this is not dependent upon nations, that this is a worldwide endeavor?
01:16:44.000I think that's going to be important once it starts to get a very high level of intelligence.
01:16:51.000In the early stages, okay, what would it hurt?
01:16:55.000If I had in my own database a central record of everything, like I'm an honest person, I'm not going to rip anyone off.
01:17:04.000But once we start to make a transition...
01:17:07.000Toward artificial general intelligence in this global decentralized network, which has component AIs from every country on the planet, like, at that point, once it's clear you're getting toward AGI, a lot of people want to step in and control this thing,
01:17:23.000you know, by law, by military might, by any means necessary.
01:17:27.000By that point, the fact that you have this open decentralized network underpinning everything, like, This gives an amazing resilience to what you're doing.
01:17:43.000You want it to be a global upsurge of creativity and mutual benefit from people all over the planet, which no powerful party can shut down even if they're afraid that it threatens their hegemony.
01:17:56.000It's very interesting because in a lot of ways it's a very elegant solution to what's an obvious problem.
01:18:24.000If you're a person running the US or China, you...
01:18:28.000Would have a different relationship than if you're a person, like I know the Prime Minister of Ethiopia, Abiy Ahmed, who has a degree in software engineering, and he loves this.
01:18:39.000But of course, Ethiopia isn't in any date.
01:18:42.000Suppressing any other countries, right?
01:18:43.000And they're not in any danger of individually, like, taking global AI hegemony, right?
01:18:48.000So for the majority of countries in the world, they like this for the same reason they like Linux, right?
01:18:55.000I mean, this is something in which they have an equal role to anybody else.
01:19:01.000And you see this among companies also, though.
01:19:04.000So a lot of big companies that we're talking to...
01:19:08.000They like the idea of this decentralized AI fabric because, I mean, if you're not Amazon, Google, Microsoft, Tencent, Facebook, so on, if you're another large corporation, you don't necessarily want all your AI and all your data To be going into one of this handful of large AI companies,
01:19:27.000you would rather have it be in a secure, decentralized platform.
01:19:32.000I mean, this is the same reason that Cisco and IBM, they run on Linux.
01:20:40.000In some way, then your business is basically controlled by this other company.
01:20:48.000So having a decentralized platform in which you're...
01:20:53.000You know, an equal participant along with everybody else is actually a much better position to be in.
01:20:59.000And I think this, I think, is why we can succeed with this plan of having this, you know, decentralized singularity net platform than this singularity studio enterprise software company which mediates between the decentralized platform and big companies.
01:21:18.000I mean, it's because most companies and governments in the world They don't want hegemony of a few large governments and corporations either.
01:21:28.000And you can see this in a lot of ways.
01:21:31.000You can see this in embrace of Linux and Ethereum by many large corporations.
01:21:37.000You can also see, in a different way, the Indian government...
01:21:43.000You know, they rejected an offer by Facebook to give free internet to all Indians, because Facebook wanted to give, like, mobile phones, it would give free internet, but only to access Facebook, right?
01:21:57.000And India is now giving, they're now creating laws that...
01:22:02.000Any internet company that collects data about Indian people has to store that data in India, which is so the Indian government can subpoena that data when they want to, right?
01:22:12.000So you're already seeing a bunch of resistance against hegemony by a few large governments or large corporations, right?
01:22:22.000By other companies and other governments.
01:22:24.000I think this is very positive and is one of the factors that can foster the growth of a decentralized AI ecosystem.
01:22:34.000Is it fair to say that the future of AI is severely dependent upon who launches it first?
01:22:43.000Like whoever, whether it's singularity net, or whether it's artificial general intelligence.
01:22:49.000The bottom line is, as a scientist, I have to say we don't know, right?
01:22:52.000It could be there's an end state that AGI will just self-organize into, almost independent of the initial condition, but we don't know.
01:23:05.000And given that we don't know, I'm operating under the, you know, the heuristic...
01:25:01.000So SingularityNet in itself is a platform that allows many different AIs to operate on it, and these AIs can offer services to anyone who requests services of the network,
01:25:16.000and they can also request and offer services among each other.
01:25:22.000So it's both just an online marketplace for AIs, much like You know, the Apple App Store or Google Play Store, but for AIs rather than phone apps.
01:25:33.000But the difference is the different AIs in here can outsource work to each other and talk to each other.
01:25:39.000And that gives a new dimension to it, right?
01:25:41.000Where you can have, we think of as a society or economy of minds, and it gives the possibility that this whole society of interacting AIs...
01:25:51.000Which are then, they're paying each other for transactions with our digital money, our cryptographic token, which is called the AGI token.
01:26:01.000So these AIs, which are paying each other and rating each other of how good they are, sending data and questions and answers to each other, can self-organize into some overall AI mind.
01:26:13.000Now, we're building this platform and then we're plugging into it To seed it a bunch of AIs of our own creation.
01:26:20.000So I've been working for 10 years on this open source AI project called OpenCog, which is oriented toward building general intelligence.
01:26:28.000And we're putting a bunch of AI agents based on the OpenCog platform.
01:26:55.000Within the larger pool of AIs on the Singularity Net can sort of serve as the general intelligence core because the OpenCog AI agents are really good at abstraction and generalization and creativity.
01:27:08.000We can put a bunch of other AIs in there that are good at highly specific intelligence.
01:27:13.000Forms of learning like predicting financial time series, curing diseases, answering people's questions, organizing your inbox.
01:27:21.000So you can have the interaction of these specialized AIs and then more general purpose, you know, abstraction and creativity-based AIs like OpenCog Agents all interacting together.
01:27:35.000And then, you know, the beauty of it is like some 15-year-old genius in Azerbaijan or the Congo can put some brilliant AI into this network.
01:27:44.000If it's really smart, it will get rated highly by the other AIs for its work helping them do their thing.
01:27:52.000Then it can get replicated over and over again across many servers.
01:27:56.000Suddenly, A, this 16-year-old kid from Azerbaijan or the Congo could become wealthy from their copies of their AI, providing services to other people's AIs.
01:28:07.000And B, the creativity in their mind is out there and is infusing this global AI network with some...
01:28:15.000Some new intellectual DNA that never would have been found by a Tencent or a Google because they're not going to hire some Congolese teenager who may have a brilliant AI idea.
01:28:31.000So this is all ongoing right now, and the term singularity that you guys are using, the way I've understood that term, correct me if I'm wrong, is that it's going to be the one innovation or one invention that essentially changes everything forever.
01:28:47.000Well, singularity isn't necessarily one invention.
01:29:33.000But he opted not to become a pundit about it because he'd rather write more science fiction books.
01:29:40.000That's interesting that a science fiction author...
01:29:42.000Ray Kurzweil, who's also a good friend of mine, I mean, Ray...
01:29:46.000I took that term and fleshed it out and did a bunch of data analytics trying to pinpoint when it would happen.
01:29:55.000But the basic concept of the technological singularity is a point in time when technological advance occurs so rapidly that to the human mind it appears almost instantaneous.
01:30:07.000Like imagine 10 new Nobel Prize winning discoveries every second or something, right?
01:30:13.000So, this is similar to the concept of the intelligence explosion that was posited by the mathematician I.J. Goode in 1965. What I.J. Goode said then, the year before I was born, was the first truly intelligent machine will be the last invention that humanity needs to make,
01:30:31.000So, this is an intelligence explosion is another term for basically the same thing as a technological technology.
01:30:38.000But it's not just about AI. AI is just probably the most powerful technology driving it.
01:30:44.000I mean, there's AI, there's nanotechnology, there's femtotechnology, which will be building things from elementary particles.
01:30:51.000I mean, there's life extension, genetic engineering, mind uploading, which is like reading the mind out of your brain and putting it into a machine.
01:31:01.000You know, there's advanced energy technologies so that...
01:31:05.000All these different things are expected to advance at around the same time, and they have many ways to boost each other, right?
01:31:12.000Because the better AI you have, your AI can then invent new ways of doing nanotech and biology.
01:31:18.000But if you invent amazing new nanotech and quantum computing, that can make your AI smarter.
01:31:22.000On the other hand, if you could crack how the human brain works and genetic engineering to upgrade human intelligence, those smarter humans could then make better AIs and nanotechnology, right?
01:31:32.000So there's so many virtuous cycles Among these different technologies, the more you advance in any of them, the more you're going to advance in all of them.
01:31:42.000And it's the coming together of all of these that's going to create, you know, radical abundance and the technological success.
01:31:51.000So that term which Werner Vinci introduced, Ray Kurzweil borrowed for his books and for the Singularity University educational program, and then we borrowed that for our Singularity Net decentralized blockchain-based AI platform and our Singularity Studio enterprise software company.
01:32:13.000Now, I want to talk to you about two parts of what you just said, one being the possibility that one day we can upload our mind or make copies of our mind.
01:32:34.000Do you think that that's a real possibility inside of our lifetime, that we can map out the human mind to the point where we can essentially recreate it?
01:32:42.000But if you do recreate it, without all the biological urbs and the human reward systems that are built in, what the fuck are we?
01:33:05.000One thing is a better computing infrastructure than we have now to host the uploaded body.
01:33:12.000And the other thing is a better scanning technology, because right now, we don't have a way to scan the molecular structure of your body without freezing you, slicing you, and scanning you, which you probably don't want done at this point in time.
01:33:28.000So, assuming both those are solved, you could then recreate in some computer simulation...
01:33:36.000You know, an accurate simulacrum of what you are, right?
01:33:43.000An accurate simulacrum, that's getting weird because the biological variability of human beings, we vary day to day.
01:33:51.000And your simulacrum would also vary day to day, so it would deviate.
01:33:55.000You would program it in to have flaws?
01:33:58.000Because we vary depending upon how much sleep we get, whether or not we're feeling sick, whether we're lonely.
01:34:03.000If your upload were an accurate copy of you, then the simulation hosting your upload would need to have an accurate simulation of the laws of biophysics and chemistry.
01:34:16.000That allow your body to, you know, evolve from one second to the next.
01:34:20.000My concern is that it's going to recognize.
01:34:21.000Your upload would change second by second just like you do, and it would diverge from you, right?
01:34:27.000So, I mean, after an hour, it will be a little different.
01:34:30.000After a year, it might have gone in a quite different direction for you.
01:34:34.000It'll probably be a monk, some super god monk living on the top of a mountain somewhere in a year.
01:34:46.000You're not talking about the potential of downloading this again into a biological...
01:34:51.000There's a lot of possibilities, right?
01:34:53.000I mean, you could upload into a Joe Rogan living in a virtual world and then just create your own fantasy universe, or you could 3D print an alternate synthetic...
01:35:06.000I mean, once you have the ability to manipulate molecules at will, the scope of possibilities becomes much greater than we're used to thinking about.
01:35:18.000My question is, do we replicate flaws?
01:37:10.000And ultimately the ramifications were not foreseen by people 500 years ago.
01:37:16.000I mean, we're going into a lot of new domains.
01:37:19.000We can't see the details of the pluses and minuses that are going to unfold.
01:37:25.000It would behoove us to simply become comfortable with radical uncertainty because otherwise we're going to confront it anyway and we're just going to be nervous.
01:38:04.000Inasmuch as we humans can know anything, it would seem, commonsensically, there's the ability to bias this in a positive rather than negative direction.
01:38:16.000We should be spending more of our attention on doing that rather than, for instance, Advertising, spying and making chocolatier chocolates and all the other things.
01:38:27.000Right, but how many people are doing that?
01:38:29.000But I mean, how many people are actually at the helm of that as opposed to how many people are working on various aspects of technology all across the planet?
01:38:55.000Lo and behold, they're amazingly useful for training neural net models, which is one among many important types of AI, right?
01:39:02.000So a large amount of the planet's resources are now getting spent on technologies that are Indirectly supporting these singularitarian technologies.
01:39:12.000So as another example, like microarrayers, they let you measure the expression level of genes, how much each gene is doing in your body at each point in time.
01:40:31.000Now, but existential angst, just when people sit and think about the pointlessness of our own existence, like we are these finite beings that are clinging to a ball that spins a thousand miles an hour, hurling through infinity, what's the point?
01:40:46.000There's a lot of that that goes around already.
01:40:48.000If we create an artificial environment that we can literally somehow or another download a version of us, and it exists in this...
01:41:00.000Blockchain-created or powered weird fucking simulation world, what would be the point of that?
01:41:18.000It's a bit personal and maybe different than many of my colleagues.
01:41:21.000What I really believe is that these advancing technologies are going to lead us to unlock many different states of consciousness and experience than Most people are currently aware of.
01:41:59.000We understand almost nothing about who and what we are, and our knowledge about the universe is extremely minuscule.
01:42:10.000I mean, if anything, I look at things from more of a Buddhist or phenomenological way, like there's sense perceptions, and then out of those sense perceptions, models arise and accumulate, including a model of the self and a model of the body,
01:42:28.000And the model of the physical world out there.
01:42:31.000And by the time you get to planets and stars and blockchains, you're building hypothetical models on top of hypothetical models.
01:42:39.000And then, by building intelligent machines and mind-uploading machines and virtual realities, we're going to radically transform You know, our whole state of consciousness,
01:42:54.000our understanding of what mind and matter are, our experience of our own selves, or even whether a self exists.
01:43:02.000And I think, ultimately, the state of consciousness of a human being like a hundred years from now, after a technological singularity, is going to bear very little resemblance to the states of consciousness we have No,
01:43:19.000we're just going to see a much wider universe than any of us now imagined to exist.
01:43:28.000Now, this is my own personal view of things.
01:43:31.000You don't have to agree with that to think the technological singularity will be...
01:43:38.000Ray Kurzweil and I agree there's going to be a technological singularity within decades at most.
01:43:46.000And Ray and I agree that if we bias technology development appropriately, we can very likely guide this to be a world of abundance and benefit for humans as well as AIs.
01:44:20.000We understand really, really little of what we are and what this world is.
01:44:26.000And this is part of my own personal quest for wanting to upgrade my brain and wanting to create artificial intelligences.
01:44:35.000It's like I've always been driven above all else by wanting to understand everything I can about the world.
01:44:40.000So, I mean, I've studied every kind of science and engineering and social science and read every kind of literature.
01:44:46.000In the end, the scope of human understanding is clearly very small, although at least we're smart enough to understand how little we understand, which I think my dog doesn't understand, how little he understands, right?
01:44:58.000And even like my 10-month-old son, he understands how little he understands, which is interesting, right?
01:45:10.000I mean, everything we think and believe now is going to seem absolutely absurd to us after there's a singularity.
01:45:16.000We're just going to look back and laugh in a warm-hearted way at all the incredibly silly things we were thinking and doing back when we were trapped in our primitive biological brains and bodies.
01:45:30.000It's stunning that that, in your opinion or your assessment, is somewhere less than a hundred years away from now.
01:45:37.000Yeah, that requires exponential thinking, right?
01:45:41.000That's hard to wrap your head around, right?
01:45:52.000It took me some time to get my parents to wrap their head around it because they're not technologists.
01:45:58.000I mean, I find if you get people to pay attention and sort of lead them through all the supporting evidence, Most people can comprehend these ideas reasonably well.
01:46:12.000Go back to computers from 1963. It's just hard to grab people's attention.
01:46:16.000And mobile phones have made a big difference.
01:46:19.000I spent a lot of time in Africa, in Addis Ababa, in Ethiopia, where we have a large AI development office.
01:46:54.000I mean, I think the word simulation is probably wrong, but yet the idea of an empirical, you know, materialist physical world is almost certainly wrong also.
01:47:09.000Well, again, if you go back to a phenomenal view, I mean, you could look at the mind as primary, and, you know, your mind is building...
01:47:22.000The world as a model, as a simple explanation of its perceptions.
01:47:27.000On the other hand, then what is the mind?
01:47:30.000The self is also a model that gets built out of its perceptions.
01:47:35.000But then, if I accept that your mind has some fundamental existence also, based on a sort of I-you feeling that you're like a mind there, our minds are working together To build each other and to build this world.
01:47:50.000And there's a whole different way of thinking about reality in terms of first and second person experience rather than these empiricist views like this is a computer simulation or something.
01:48:04.000Right, but you still agree that this is a physical reality that we exist in, or do you not?
01:48:12.000Is it your interpretation of this physical reality?
01:48:14.000If you look in modern physics even, quantum mechanics, there's something called the relational interpretation of quantum mechanics, which says that there's no sense in thinking about an observed entity.
01:48:27.000You should only think about an observed, comma, observer pair.
01:48:31.000Like there's no sense to think about some thing except from the perspective of some observer.
01:48:38.000So that's even true within our best current theory of modern physics as induced from empirical observations.
01:48:48.000But in a pragmatic sense, you know, if you take a plane and fly to China, you actually land in China.
01:49:07.000Well, if you think about it, like, how do you know that you're not a brain floating in a vet somewhere which is being fed illusions by a certain evil scientist and two seconds from now This simulated world disappears and you realize you're just a brain in a vat again.
01:51:54.000He had the same idea, like some post-singularity intelligence, which actually was living outside of time somehow, is reaching back and putting into his brain the idea of how to...
01:52:32.0002007 or so, this guy came to Washington, where I was living then, and he brought my friend Hugo DeGarris, another crazy AI researcher with him, and he's like, the singularity is going to happen in 2012, because Terence McKenna said so,
01:52:48.000and we need to be sure it's a good singularity.
01:52:52.000So, you can't move to China, then it will be a bad singularity.
01:52:57.000So, we have to get the U.S. government...
01:52:59.000To give billions of dollars to your research to guarantee that the singularity in 2012 is a good singularity, right?
01:53:07.000So he led us around to meet with these generals and various high hoo-has in D.C. to get them to fund Hugo de Garra says in my AI research to guarantee I wouldn't move to China and Hugo wouldn't move to China so the U.S. would create a positive singularity.
01:55:50.000You don't know what the fuck you're saying.
01:55:52.000And you become a prisoner to your words in a lot of ways.
01:55:56.000You get locked up in this idea of expressing this thought that may or may not be viable.
01:56:02.000I'm not sure that he was after empirical truth in the same sense that, say, Ray Kurzweil is.
01:56:09.000When Ray is saying, we're going to get human-level AI in 2029, and then, you know, massively superhuman AI in a singularity in 2045, I mean, Ray is very literal.
01:56:35.000So you have to take that in a poetic sense rather than in a literal sense.
01:56:41.000And yeah, I think it's very interesting...
01:56:44.000To go back and forth between the symbolic and poetic domain and the concrete science and engineering domain.
01:56:53.000But it's also valuable to be able to draw that distinction, right?
01:56:58.000Because you can draw a lot of insight from the kind of thinking Terence McKenna was doing.
01:57:04.000And certainly, if you explore psychedelics, you can gain a lot of insights into how the mind and universe work.
01:57:11.000But then when you put on your science and engineering mindset, you want to be rigorous about which insights do you take and which ones do you throw out, and ultimately you want to proceed on the basis of what works and what doesn't, right?
01:57:25.000I mean, Dennis was pretty strong on that, and Terence was a bit less in that empirical direction.
01:57:31.000Well, Dennis is actually a career scientist.
01:57:48.000Unfortunately, due to the illegal nature of these things, it's a little hard to pin down.
01:57:56.000Before the recent generation of people going into AI because it was a way to make money, the AI field was incredibly full of really, really interesting people and deep thinkers about the mind.
01:58:09.000And in the last few years, of course, AI has replaced business school as what your grandma wants you to do to have a good career.
01:58:17.000So, I mean, you're getting a lot of people into AI just because it's...
01:58:27.000Because in our generation, AI was not what your grandma wanted you to do so as to be able to buy a nice house and support a family, right?
01:58:37.000So you got into it because you really were curious about how the mind works.
01:58:41.000And of course, many people played with psychedelics because they were curious about...
01:58:48.000You know, what it was teaching them about how their mind works.
01:58:53.000I had a nice long conversation with Ray Kurzweil, and we talked for about an hour and a half, and it was for this sci-fi show that I was doing at the time.
02:01:29.000You'll probably have a QPU, quantum processing unit, in your phone in like 10 to 20 years or something, right?
02:01:36.000So that might throw off the 2045 date, but in a historical sense, it doesn't change the picture.
02:01:44.000I've got a bunch of research sitting on my hard drive on how we improve OpenCog's AI using quantum computers once we have better quantum computers, right?
02:01:54.000There could be other things like that, which are technical roadblocks that we're not seeing now, but I really doubt those are going to delay things by more than a decade or two or something.
02:02:06.000On the other hand, things could also go faster than Ray's prediction, which is what I'm pushing towards.
02:02:40.000I mean, it can use many robots as user interfaces, but the same AI could control many different robots, actually, and many other sensors and systems besides robots.
02:02:50.000I mean, I think the human-like form factor, like we have with Sophia and our other Hansen robots, the human-like form factor is really valuable as a tool for allowing the cloud-based AI mind to To, you know, engage with humans and to learn human cultures and values.
02:03:06.000Because, I mean, getting back to what we were discussing at the beginning of this chat, you know, the best way to get human values and culture into the AI is for humans and AIs to enter into many shared, you know, like social, emotional, embodied situations together.
02:03:20.000So having a human-like embodiment for the AI is important for that.
02:03:26.000Like the AI can look you in the eye, it can share your facial expressions, it can bond with you.
02:03:31.000It can see the way you react when you see like a sick person by the side of the road or something, right?
02:03:36.000And, you know, it can see you ask the AI to give the homeless person $20 or something.
02:03:43.000I mean, the AI understands what money is and understands what that action means.
02:03:48.000Interacting with an AI in human-like form is going to be valuable as a learning mechanism for the AI and as a learning mechanism for people to get more comfortable with AIs.
02:03:59.000But I mean, ultimately, one advantage of being, you know, a digital mind is you don't have to be wedded to any particular embodiment.
02:04:06.000The AI can go between many different bodies and it can transfer knowledge between the many different bodies that it's occupied.
02:04:13.000Well, that's the real concern that the people that are...
02:04:17.000That have this dystopian view of artificial intelligence have is that AI may already exist and it's just sitting there waiting.
02:07:40.000Was it freaky watching that, though, with the name Ava?
02:07:44.000The thing is, the moral of that movie is just, if a sociopath raises a robot with an abusive interaction, it may come out to be a sociopath or a psychopath.
02:09:10.000Well, right now we can load it in English mode, Chinese mode, or Russian mode.
02:09:15.000And there's sort of different software packages.
02:09:18.000And we also use her sometimes to experiment with the OpenCog system and SingularityNet.
02:09:24.000So we can use the robot as a research platform for exploring some of our more advanced AI tools.
02:09:31.000And then there's a simpler chatbot software, which is used for appearances like that one.
02:09:36.000And in the next year, we want to roll out more of our advanced research software from OpenCog and SingularityNet, roll out more of that inside these robots, which is one among many applications we're looking at with our SingularityNet platform.
02:09:52.000I want to get you back in here in like a year and find out where everything is.
02:09:56.000Because I feel like we need someone like you to like...
02:10:19.000I mean, we think about the singularity like it's going to be some huge, like, physical event and suddenly everything turns purple and it's covered with diamonds or something, right?
02:10:31.000But, I mean, there's a lot of ways something like this could unfold.
02:10:34.000So, like, imagine that with our singularity net decentralized AI network, you know, we get...
02:10:41.000An AI that's smarter than humans and can create, you know, a new scientific discovery of the Nobel Prize level every minute or something, that doesn't mean this AI is going to immediately, like, refactor all matter into images of Buckethead or do something random,
02:11:30.000And we're using the Sophia robot as a meditation assistant.
02:11:35.000So we're using Sophia to help people get into deep, like, meditative trance states and help them, you know, breathe deeply and achieve more positive state of being.
02:11:48.000And part of the goal there is to help people.
02:11:50.000Part of the goal is as the AI gets more and more intelligent, you're sort of getting the AI locked into a very positive, reflective, and compassionate state.
02:12:01.000I think there's a lot of things in the human psyche and evolutionary history that hold us back from being optimally compassionate.
02:12:09.000And that if we create the AI in the right way, it will be not only much more intelligent, but much more compassionate than human beings are.
02:12:41.000If we can't create something that's not only more intelligent but more wise and compassionate than we are, we're probably going to destroy ourselves by some method or another.
02:12:51.000I mean, with something like Donald Trump becoming president, you see what happens when this primitive hindbrain and when our unchecked mammalian emotions of anger and status-seeking and ego and rage and lust...
02:13:08.000When these things are controlling these highly advanced technologies, this is not going to come to a good end.
02:13:16.000So we want compassionate general intelligences, and this is what we should be orienting ourselves toward.
02:13:37.000Because you need to work with the establishment rather than Overthrowing it, which isn't going to be viable.
02:13:57.000Then we're creating these robots like Sophia, which will be mass manufactured in the next couple of years, roll these out as service robots everywhere around the world to interact with people, providing valuable services in homes and offices, but also interacting with people in a loving and compassionate way.
02:14:20.000Now, because we don't actually know if it's going to be years or decades before we get to this singularity, and we want to be as sure as we can that when we get there, it happens in a beneficial way for everyone, right?
02:14:32.000And things like robots, blockchain, and AI learning algorithms are tools toward that end.
02:14:39.000Well, Ben, I appreciate your optimism.
02:14:41.000I appreciate coming in here and explaining all this stuff for us, and I appreciate all your work, man.
02:14:50.000It's a really fun, wide-ranging conversation.
02:14:52.000So, yeah, it would be great to come back next year and update you on the state of the singularity.
02:14:57.000Yeah, let's try to schedule it once a year, and just by the time you come, maybe, who knows, a year from now, the world might be a totally different place.