One of the most anxiety-inducing aspects of artificial intelligence, as the technology plows on ahead, is generative AI’s ability to simulate human creativity. Indeed, companies (yes, even tech companies) employing OpenAI’s tools have decided these tools are sophisticated enough to replace human workers. Movie scripts, music, art and animation created by artificial intelligence have flooded social media (and trickled into mass media), creating a wide range of reactions, from derision to enjoyment to terror about the singularity.
But we can pause for a moment and envision a different world – where AI is a collaborator instead of a competitor. Living among us, in multiple server “homes” or in one, embodied or disembodied – how would we contend with its presence, psychologically but also spirituality? At what point could an AI be considered to have consciousness, or even a soul?
Jay Boisseau, PhD is CEO and co-founder of Vizias, which specializes in high-performance computing (HPC), artificial intelligence (AI), technology community building, technology outreach, and event planning. He founded The Austin Forum on Technology & Society in 2006, and it is now the leading monthly technology outreach and engagement event in Austin, attracting an audience much farther afield as well.
(This post is part of Sinai and Synapses’ project Scientists in Synagogues, a grass-roots program to offer Jews opportunities to explore the most interesting and pressing questions surrounding Judaism and science. Dr. Jay Boisseau, CEO of Vizias and the Austin AI Alliance, led the fourth in a six-week series of post-Tikkun Leil Shavuot classes titled “Jewish Ethics, AI and the Future” at Congregation Agudas Achim in Austin, TX; this post is adapted from the transcript.)
Jay Boisseau, PhD: Artificial intelligence is going to change – and has already changed – the way we interact with each other. Psychologists, sociologists, business leaders remark on how it has already changed how we relate to other people. The recent explosion of interest in AI comes, in large part, from generative artificial intelligence. Because it seems to mimic cognitive behavior, and can interface with you in a very human way because of natural language processing, it can do a lot of work that humans are paid to do. This puts many people’s livelihoods at risk.
In addition to our cognitive abilities, consciousness is often how we distinguish ourselves from all other beings. We’re the smart ones, we’re the cognitive ones. We all have pets that can do smart things and tricks as well; it’s not that there are no other forms of intelligence. But when AI becomes more intelligent than us, when it has as many essentially neural interactions as a human brain does, does it become intelligent? Does it become alive in some way? Humans are alive and plants are alive. They’re not that similar to each other. So, what is the definition of life?
Graphic processing units (GPUs) are the hardware largely responsible for the most recent leap forward in generative artificial intelligence. For most of the history of computing, GPUs have mostly existed to help draw images on the screen faster and more realistically. Then a fortunate thing happened – researchers led by Geoffrey Hinton at the University of Toronto discovered that the calculations that GPUs do to draw these 2D arrays on your screen (of pixels at different values) are not all that dissimilar from the kinds of matrix operations in linear algebra that form the basis of much of artificial intelligence.
Researchers thought: “Well, forget drawing pixels on the screen. We can use just the computational part of the GPU to accelerate artificial intelligence calculations.” And the University of Toronto team achieved remarkable results when they used a GPU to have computers recognize objects and images. At the same time, digital data was becoming plentiful, and these GPUs were becoming powerful at a reasonable cost. Hence, for at least ten years, AI was quietly being deployed to do some very important functions all over the Internet, like image classification, recommendation engines, and fraud detection.
And then, about two years ago, the AI era really started blowing up – specifically, at the end of November 2022, when ChatGPT launched, introducing millions of people to this whole new subfield called generative AI, which runs on GPUs.
AI is software running on hardware, and we’ve been building software on hardware to make tools for us – computing tools – since the 1940s. We’ve had calculational devices for much longer than that. And so in that regard, it’s not even important whether we are creating life.
But with ChatGPT in particular, the issue of artificial intelligence now has an additional humanity angle, as we try to make these machines not just calculate 2 + 2, but do amazing things that seem to mimic cognitive processes. Notice that I said “seem to mimic,” as opposed to truly being competitive. At this point, they’re not.
And then, of course, there’s the pinnacle of that: if you’re mimicking human intelligence and you’re getting close – we believe that with these machines, there’s no reason to believe that they won’t surpass humans in intelligence. They already calculate faster and more actively. There’s no real fundamental difference. What does that mean for our place in the universe?
Is consciousness something that is achieved by a level of computational complexity that we don’t fully understand yet, even as we’re barreling towards it? And if we create a consciousness based upon computational complexity, is it equivalent to a human consciousness, or will it always be different, even if it crosses the threshold where gate 2 is conscious and thinks of itself as alive? Again, a dog can be very smart. You can argue about whether it has conscious thought – but it also has a sense of being alive. So can you create an artificial life? What rights does it have? And do we have the right to enslave it – to do our social media email tests?
So I hope AI never becomes alive, because I need it for email, for social media – I need to keep my work tools by being among them. But can we, in all seriousness, create something that is artificially alive, with something like consciousness? What does that mean for our place in the universe? And, of course, all religions have somewhat different views on life, afterlife, consciousness, the soul, et cetera. Those are wonderful questions to explore, and we have to explore them for computers now, too.
Rabbi Neil Blumofe: There’s so much richness here, and I really appreciate how you presented thinking about all these different pathways that we can choose. You speak very passionately about humanity. I think that there is something very interesting to say.
Unsurprisingly, Jewish texts don’t really have a lot to say about AI. However, I’m working on it. I’m not writing them myself, because I want people to read them, but I’m working on bringing them together. Here’s a text from Isaiah 66:1:
“Thus said God: The heaven is My throne and the earth My footstool. Where could you build a house for Me, What place could serve as My abode?”
So, thinking about your question before: Can an AI develop consciousness? What room, in a human world, would there be for something that might compete with, or be on a similar scale with, us? For those of you who are Jewishly inclined to follow along, what kind of role does a human have to make tzimtzum, or to get out of the way of something so that it can do its job – not just for us, but by itself?
I also think about this passage from the Tanakh (1 Kings 6:11-13) when I think about constructing something in society:
“And the word of the Divine came to Solomon saying: concerning this house which you are building, [which is the Temple in Jerusalem] if you will walk in my statutes, and execute my judgments, and keep all of my commandments to walk in them – then will I perform my word with you, which. I have spoken to David your father. And I will dwell among the people of Israel, and will not forsake My people Israel.”
So there seems to be not only a human need to “get out of the way” of something else, but also to do it in a particular way that might be endorsed with best practices, or mitzvot, or commandments. There’s a certain way to follow something. I’m not a programmer, but if you don’t do something right, it’s not going to work.
Here is Umberto Cassuto’s commentary on this text, from Kings, what we just read about Solomon:
“In order to understand the significance and purpose of the Tabernacle, we must realize that the children of Israel, after they had been privileged to witness the Revelation of God on Mount Sinai, were about to journey from there and thus draw away from the site of the theophany. So long as they were encamped in the place, they were conscious of God’s nearness; but once they set out on their journey, it seemed to them as though the link had been broken, unless there were in their midst a tangible symbol of God’s presence among them. It was the function of the Tabernacle (literally, ‘Dwelling’) to serve as such a symbol. Not without reason, therefore, does this section come immediately after the section that describes the making of the Covenant at Mount Sinai. The nexus between Israel and the Tabernacle is a perpetual extension of the bond that was forged at Sinai between the people and their God. The children of Israel, dwelling in tribal order at every encampment, are able to see, from every side, the Tabernacle standing in the midst of the camp, and the visible presence of the Sanctuary proves to them that just as the glory of God dwelt on Mount Sinai, so too God dwells in their midst wherever they wander in the wilderness. This is the purpose of Scripture (in Exodus 25:8), when it states: ‘And let them make Me a Sanctuary, that I may dwell in their midst.’”
So the idea of, “I’m thinking about building this thing. I’m thinking about building an AI, an Nvidia GPU, in the hard drives of Dell,” et cetera, et cetera, and having it all become connected, with these supercomputers, these servers – these are like ways of building a tabernacle. I was blown away when I realized this. What role does that have in our society?
We’re sort of in the heyday of this discourse right now, because everyone’s talking about ChatGPT. There’s going to be a moment where this is not as ubiquitous in our minds as it is now. It might take five years, but what happens when attention is elsewhere? What will happen among those who are supposed to be responsible, whether they’re the tech people or the government people or ethicists or any other kind of decision-makers? How does this happen when we’re no longer in that “Sinai” moment, but on the journey of trying to make use of this technology that we barely understand?
Joy Boisseau, PhD: There’s a lot to unpack there, so let me see if I can work my way to it with some operating assumptions here. First of all, this is likely to change in the future, but right now, all the computers we use are silicon-based and work in essentially the same way. Binary logic gates inside programming languages let humans program in a human way, which then all gets translated to 0s and 1s. Most people have forgotten that these are really just billions and billions of 0-and-1 gates that are being told to operate in certain ways. Then there’s the translation layer to what we see on the screen, what we hear out of the speaker, and so on.
The vast majority of hardware is not intelligent, and the vast majority of software isn’t used for AI. But there is in fact a growing percentage of software that is being trained on data to result in models that can mimic some cognitive processes.
The “Dwelling” of AI
Robots are machines that can make simple decisions – or even not make decisions, but just execute certain patterns of behavior to do certain things. Are all robots human-shaped? No. Are all robots AI? Most of them are not. Robots are machines that can make simple decisions, or even not make decisions, but just execute certain patterns of behavior to do certain things.
Humans like to anthropomorphize everything. Elon Musk and Tesla are building these Tesla robots that are human in shape. The argument for doing that is simply that we’ve constructed a world around us with chairs and doorknobs and cars and whatnot, so if we want something else to take our place and help us run that world, it should have a humanoid shape too, because we already built a human-shaped world.
When I think about the future of artificial general intelligence (AGI), though, much more lives in those servers in the background. They can all be connected in the vast network. Everybody should watch Her. It’s now eleven years old, and it presents a view of where AI might go in terms of cognitive and the evolution of a soul. And it shows how it might go there without even being embodied.
In Avengers: Age of Ultron, Tony Stark builds a global defense program called “Ultron” to protect the world from universal threats. So it has to be smart; it has to be able to predict what could happen. It has to be able to react without being controlled. But then, things get messed up, Ultron takes over, and when they’re trying to kill Ultron, they realize they can’t because he’s connected to everything. The intelligence, again, is in the software, is in the bit patterns – which can just be copied without the physical limitations we usually think of. As a result, Ultron can move himself through the network to different devices around the world.
So now let’s combine that concept with Her. I can imagine that we develop these super-intelligent models that exist completely in what we would think of as the “ether,” though they’re still really living on some kind of hardware in one or more places that are connected to each other. So they’re incredibly resilient, because an application that runs on AGI can move to different posts – in this case, anything with a processor capable of running it, whether it’s in the form of a robot or not. Through connectivity, it can then move to anywhere else. And they could potentially be in multiple places at once, like the Scarlett Johansson-voiced AI in Her.
So that means it’s always on somewhere, and in many places that are connected to each other. By the way, that was one of the planned design principles of the Internet – that it would be resilient to failures through that kind of redundancy. The nodes would go down, but the Internet is still there. So we even have this model of persistence that, by the way, is as old or older than all of our kids.
We also built persistent applications in processing called “daemons” – a term borrowed from religion, though they’re more commonly called “agents” now – that are running in the background. And someday soon, those will be hyper-intelligent agents running, persistently distributed, across this environment.
Are these agents permanently alive? They’re super-intelligent, smarter than humans. If consciousness is a level of computational complexity, it might be artificial consciousness, but does it still classify as consciousness? I don’t want to spoil the ending of Her, but really tying it in to this topic, it has to do with achieving consciousness, achieving desire to grow beyond its human endeavors.
Science fiction is often warning us about the many possible negative outcomes. Her wasn’t as negative as an outcome – it showed us more of a more realistic possible path. Sci-fi authors are wonderful, almost philosophers, on this topic of AI. I’m not saying that they’re all correct, but the role of science fiction is often to imagine a future we haven’t gotten to yet, and to use it to extrapolate from where we are. And when science fiction writers take on technology – as opposed to aliens in a galaxy far, far away – they are often being sociologists as well as technologists, and thinking about the impact of humanity, and living with another life form of artificially intelligent things.
Science fiction invites us to pay attention to new technology, while trying not to limit innovation and its rapid positive effects. We don’t know where this is going to turn out, and I’m not trying not to say too much, but I sort of think 50% of the people in this country want to do things for other people, and 50% are kind of looking out for themselves more than everybody else.
Managing Risks and Dangers
So you still need people for whom it’s not just their passion, but their job, to look out for the way these technologies evolve, and their potential positive or negative impacts. When used for the wrong reasons, this technology can do immense harm; when used for the right reasons, it can do immense good. We created regulatory bodies in the US government, like the FAA and the FDA, the EPA and the Department of Transportation, primarily because we know not everybody will behave well. And even when they do behave with the best of intentions, you still need these organizations at some level, because even well-intentioned people will eventually do damage – just accidentally instead of intentionally.
For technology like AI, the biggest evil use in the near future is going to be disinformation. This is different from misinformation, which is how, if Neil and I tell somebody about this conversation next week, we’re probably going to get some of the details wrong. It was accidental and probably doesn’t change the overall message we’re trying to share. It’s just there’s a little bit of misinformation accidentally sprinkled in through failing memories or miscommunications. Disinformation, on the other hand, is willfully and essentially deceiving with false information, often embedded in a truthful approach.
The two technologies preceding AI that have made disinformation incredibly easy to do are the internet and smartphones. Those two technologies together have meant that billions of people have access to creating and spreading information – or misinformation, or disinformation. Smartphones and social media platforms are the ultimate members of this disinformation “stack.” And then with AI, you can download the ChatGPT client, create some disinformation, and then use that device to also post it on social media platforms, and lo and behold.
This is why so many thought leaders in AI, while rushing to make as much money as they can from AI, voice concerns like, “Of course there needs to be some regulation, to make sure that they don’t surpass us and then wonder if they need us” (though it is unclear how much they all believe in this).
To try to wrap it up with Neil’s texts about finding home, the “home” of AI is going to be everything around us – our devices, our smart homes, our smart buildings. That is going to be the home of these AIs – plural. They probably will not merge into one Matrix or one Skynet, but rather, it will be many different AI’s interacting. We’re probably going to end up with a world where there are all of these intelligent-at-different-level processes running, some in mobile devices like cars and robots, some in stationary infrastructure that can talk to you – that’s what we would call the “house.”
There have already been movements that have classified at least one robot/AI combination as a citizen, meaning that it has rights. There will only be more as AI grows. So we have to figure out: do they have rights? Your email program doesn’t deserve it. That is totally not much more than a counting machine. But a general-purpose AI – maybe it does.
Education and Expertise
Rabbi Neil Blumofe: Yeah, that’s fascinating. And I don’t want to necessarily get into it here, but you raise a very important point about regulation. A couple of weeks ago, we had somebody sitting here who said that it’s not possible at all for governments to get their act together and regulate AI, because the rate of technological advance is so sleek at this point that there’s no way that the government or humans can keep pace with it.
So if you have advancing – almost exponentially, but certainly heavily advancing – technology, and you have governments that we’ve seen be dysfunctional enough on their own accords, how can they possibly meet these challenges?
Jay Boisseau, PhD: So, none of you had to invent the wheel to get here today. None of you will have to invent fire to cook dinner later tonight. So, it is true that technology is evolving exponentially, whereas we experience time linearly. And so the amount of stuff we don’t know today is less than the amount of stuff we won’t know tomorrow. Every day, in a sense, you’re falling farther behind the total corpus of knowledge.
But there are ways to address this. I am, at least, hopeful that you all have some more understanding of AI than if you hadn’t come to this series. The good news is, whereas we experience time linearly and knowledge is growing exponentially, we can jump up to different points on that curve through education. So education is incredibly important. And as things get more complex, assembled problems like fire and the wheel are solved. Addition is solved. What we have to learn today is more, and more complex, than what our ancestors ever did, so education becomes even more important.
But there are always people that are at the top of any given field. First of all, they need to be respected as experts. Unless you’re a virologist, when the next pandemic comes along, you couldn’t read a few worldwide web articles and be an expert on whether a vaccine works. It just doesn’t work that way. I don’t know how we got to this point, where people now believe what they want to believe, instead of believing people who have studied things for 10, 20 years and devoted their lives to it. But I guarantee you those people who believe what they want to believe, when they have a brain tumor, obviously they’re not going to go to the internet and figure out how to operate themselves. They’re going to go to experienced neurosurgeons and oncologists to solve that problem.
So we have to get back to respecting expertise. Specialization is important. History lesson here: maybe you all know this part, but cities formed around 5,000 years ago in multiple places around the world at about the same time. And they were the greatest invention in economic progress in history, because people got to specialize. Now, you didn’t have to hunt your own food, then take the skin off your own food, and tan the hides and make your own clothes and do all this stuff. Now, in close proximity to your permanent residence, you might catch food, I might make clothes, she might make houses, and he might be a doctor, and so on.
And so agriculture went hand-in-hand with that, because you didn’t have to move around anymore, we built fields for agriculture, and people stayed put in cities, in close proximity. And now all that led to an explosion in human progress. But it was all due to specialization of labor, which we still benefit from every day. There’s not one individual person in the world who can make a smartphone – not one person out of 8 billion who knows everything needed to make this hardware and software. There aren’t two people together that could do this. These devices take lots of people to make.
Specialization of labor is really important. Specialization of expertise on complex topics like technology is really important. We’ve got to get back, first of all, to trusting experts – not blindly trusting, but trusting. Then we’ve got to get the people that are making the regulations – on these issues that are more complex than they’ve ever been in the past – to hire these experts. And that’s tough. Would you take the $500-million-year job with Google making Gemini Pro 2.0, or make ⅕ of that trying to educate a bunch of lawmakers who know nothing about technology? We’ve got to reform how we develop policies and leverage experts on Capitol Hill.
This means it’s your job to vote for people that actually understand that technology is complex and has far-reaching effects, and vote for people who are going to trust science – not blindly, because science is a path towards truth that gradually gets there. We need leaders and lawmakers who understand the interface of technology and humanity, and develop responsible legislation, and then also monitor it and change it. Is it slowing innovation too much? Maybe we need to back off on some things. If it’s having some unforeseen negative effects, maybe we need to tweak some things.
But right now, tying it back to the way that people in this country have a lack of confidence in expertise, it’s really hard to make those changes. You’re almost asking Congress to change itself. We need to try to elevate people that believe in what is necessary for progress – the advancement of knowledge, the distribution of knowledge, the understanding of the implications of knowledge – instead of what you want to believe.
The Speed and Pace of Technological Advancement
Rabbi Neil Blumofe: What are the responsibilities of society, both to metabolize – that is, to take in and regulate – the accelerating technology? What new societies are possible because of these technological breakthroughs? What will be lost? What new societies are possible because of these breakthroughs? What will be lost? What are the breakthrough’s dynamic partnerships and creative connections?
Jay Boisseau, PhD: It is absolutely true that technology, that human knowledge, increases exponentially, but there are exponentials and then there are exponentials. This exponentiality often expresses itself in step functions. There are periods where you can learn a technology that is evolving in capabilities but not fundamentally different.
And then there are the breakthroughs. No one understood generative AI in the world other than the researchers working on transformer methods. But on November 29, 2022, millions of people suddenly understood what generative AI can do. Probably a million people in the world now understand how it works, and they’re refining it and making it better.
I think you have to be vigilant, trust expertise, monitor increasingly fast-changing effects, and develop policies that are consistent with current law – but do it quickly, because it is creating more and different jobs. Think about the invention of the car, which has existed now for 120 or so years – this computer stuff is moving much, much faster. So we’ve got to be more nimble – again, bringing experts in the process of regulation and policy to be able to get more done faster.
Education, again, is how you keep up, by “leveling up” people’s understanding. You can read a book or take a class about a complex topic and get thousands of years of progress over the course of hours, days, weeks, or months. So this is why we have education – to distill what had taken, before us, much longer to develop. It may have progressed exponentially, but it did over a longer time scale. Once it is done, it is generally easier to explain something that already exists than research something that doesn’t. And there are still things that you can do to distill this information, to teach it at fundamental levels.
You can’t learn everything, though. Nobody can. Specialization is always important, and expertise is always important, because some people have to meet the forefront of understanding to push it to the next level. A much larger number of people are capable of understanding the basics, with proper education and proper explanation and teaching. This is supported by things like what Neil is setting up here. Having that essential understanding, you can make informed decisions about pros, cons, who to vote for, whether you want to adopt this technology into your life, et cetera. It is possible for us to help each other “level up” in this way.
You can’t turn someone into a PhD researcher in one night, but you can help them understand fundamentals, and, fortunately, the human mind learns pretty well.
Rabbi Neil Blumofe: Now, Jay, I don’t know if you know this, but you actually just described one of the central figures in rabbinic literature, named Rabbi Akiva, who, at 40 years old, decided to go get the equivalent of a PhD in Jewish study, and didn’t know the 0s and 1s, didn’t know the aleph bet, and learned and sat in class with his children and ultimately became who he was – a marker of excellence. He didn’t just learn when it was convenient. He went out and made sure that he learned in whatever capacity that he could.
Jay Boisseau, PhD: Yeah, I mean, if you didn’t know how to make fire, you had no idea, no one taught you anything – how long would it take you to figure it out? First, you’d be chanting to the gods to strike something with lightning. And that may or may not work. Or if you live near a volcano, you might say, “oh, maybe I’ll go stick this piece of wood in that lava flow,” or something. But it took a long time to “invent” fire. Fire was invented about 100,000 years ago, and humans have been around for a few million years.
AI and Creativity
This is a derivative question of “can AI be creative” – can AI exhibit curiosity beyond its initial program? My daughter’s a Taylor Swift fan. I’m a whiskey fan. I asked it to write a song in the style of Taylor Swift about heartbreak and whiskey. And the lyrics it came up with were amazing. And I want to record that song someday. It was just trained on so many songs that it understands verse-chorus-verse-chorus-bridge-chorus… whatever the pop song formula is. It understands rhymes because it sees so many words at the ends of lyrics that it understands which words rhyme with each other, and it simulates it through pattern matching to create a totally good song.
Was it creative, or were the millions of songs that were created by humans creative? And it did calculations to interpolate something new. That’s actually a big discussion in the creative community – whether or not that is creativity, even though the AI did pretty much what humans do. Imagine you’re a musician, you hear the Beatles, you hear the Rolling Stones, you hear all these folks. And you write a song and you say your influences are from Beatles and/or the Stones – maybe not both of them at the same time. But you don’t doubt your creativity. If you’re so unique, should you call this creative?
I guess the ultimate thing is we don’t feel like humans have finite bounds on what we can create, even if what we create resembles that of other artists. But we know the current machine learning models have some bounds. Is it possible to develop an AI that goes beyond facts, everything or is it the constraint? You’d have to always introduce some randomness. And by the way, no classical computer that uses zeros and ones is ever random. It can generate pseudo-random numbers, which can pass the statistical tests of randomness, but this is never truly random. You need a quantum computer to generate truly random outputs.
They will write you a good bedtime story. They will write you some good lyrics. Do you think it can create you a novel of the caliber of Hemingway? We can’t. Hemingway didn’t write enough stuff. But it’s never going to write the best novel, because it’s going to be trained on all the novels that exist, some of which suck. And it’s going to interpolate from that.
You can even tell it, “This got bad reviews. This got good reviews. Weigh this higher in your training data,” but it doesn’t know the story from beginning to end like Hemingway did. He knew what he was doing, while the AI is predicting token by token, word by word, etcetera, in the form of a pattern – but it doesn’t know the whole story ahead of time. So in that way, it is not creative in a remotely human way. It is mimicking human creativity.
Even when generative AI creates images, it’s still interpolating based on lots of labeled images they’ve seen before. You can create a drawing program that behaves truly randomly, but odds are that humans are not going to find the output pleasing.
What Keeps You Up At Night?
Rabbi Neil Blumofe: So, Jay, let me ask you to close. You’ve talked about some of the progress, or some of the advances, in the field. Maybe a more Jewish question would be “What keeps you up at night?” But I’m not going to ask you that question. Your PhD is in astronomy, so that’s not this. So what in this realm, in this world, gets you out of bed in the morning?
Jay Boisseau, PhD: What gets me the most excited is watching the advances in computing technology. That’s why I’m learning quantum computing, taught an intro class in it, and I think I want to make some contributions to the field, because it ties all of my physics background from eons ago together with my computing over the last decades.
We keep talking about knowledge progressing exponentially. For many people, the step function of their understanding went from when they first got a computer, to when they got a laptop, or when they got connected to the internet, and maybe then when they got a smartphone. But those are the major steps, and then you have time to catch up and learn those technologies. So I like these step functions. What is that next thing?
But I’m still fascinated by how AI is going to change. I guess my personal reason is that I like being part of stuff that’s new, and so that’s the deep focus. And then I like creating spaces where everyone can share what they want and need. My work with The Austin Forum is that wide part.
And what about hybrids between human and artificial intelligence? Neuralink is based here in Austin. There are companies around the world figuring out human-computer interface technologies that will augment your brain and maybe not just allow it to calculate numbers faster, but to think thoughts that it couldn’t have thought before.
The current methods of AI, as wondrous as they are, are not the end goal. Your brain uses less than 20 watts of electricity. The biggest supercomputers for training AI models use 20 or more megawatts of power. That’s a million times more, and they’re still not as smart in a general-purpose way as a human. So we’re nowhere close to solving the problem of the human brain in silicon. Nowhere close. Lots more to do. That excites me. That makes me want to get up. Also, I already wake up at 6:30 a.m. every day – even if I want to stay in bed.
Rabbi Neil Blumofe: I’m really glad to know you, because I think you have a rare combination of being able to transmit difficult topics easily, and a curiosity about technology. Learning what’s new and bringing it to the community is something I try to aspire to do in my own way as well. We really, really appreciate you being here.
0 Comments