Is AI Optimism In the Air?

Today on the AI Daily Brief, are we seeing the return of AI optimism? Before that in the headlines why one AI Leader Thinks AI is Dumber Than a Cat the AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show. Notes. Welcome back to the AI Daily Brief Headlines edition. All the daily AI news you need in around five minutes. New technologies always come with their fair share of controversy. Sometimes there are new behavior patterns that older generations think are going to be destructive among younger generations. Sometimes the concern is more around jobs in the economy and what the impact of new automation technology will be. With AI, we have all of that, but we also have this bigger question of whether this technology will accidentally at some point end the world. For that reason, as you can imagine, the debates around AI can be somewhat more contentious than usual. Last week, one of the leading voices in the AI safety movement who is pushing for more caution in how we handle and progress with regards to AI, Jeffrey Hinton, won a Nobel Prize for physics. He used the occasion to hope that the awarding of the prize would make people take his warnings more seriously. However, one of Hinton's fellow Turing Prize award winners from back in 2018, Yann LeCun, has a very different take on this question. As the Wall Street Journal put it in an article this weekend, this AI pioneer thinks AI is dumber than a cat. Jan LeCun, an NYU professor and senior researcher at Meta, says warnings about the technology's existential peril are quote, complete bs. Now, this is of course not a new position for Jan. Back in May, he tweeted, it seems to me that before urgently figuring out how to control AI systems much smarter than us, we need to have the beginning of a hint of a design for a system smarter than a house cat. Such a sense of urgency reveals an extremely distorted view of reality. Now the context for this post is former head of super alignment at OpenAI, Jan Leakey. Stepping away from that post, Jan Lecun it's as if someone had said in 1925, we urgently need to figure out how to control aircrafts that can transport hundreds of passengers at near the speed of sound over the oceans. It would have been difficult to make long haul passenger jets safe before the turbojet was invented and before any aircraft had crossed the Atlantic nonstop. Yet we can now fly halfway around the world on twin engine jets in complete safety. It didn't require some sort of magical recipe for safety. It took decades of careful engineering and iterative refinements the process will be similar for intelligent systems. It will take years for them to get as smart as cats, and more years to get as smart as humans, let alone smarter. Now, the big point here is where Jan says, don't confuse the superhuman knowledge accumulation and retrieval abilities of current LLMs with actual intelligence. And this is really the nexus of his point. Back in the Journal, LeCun thinks that today's AI models, while useful, are far from rivaling the intelligence of our pets, let alone us. When I ask whether we should be afraid that AIs will soon grow so powerful that they pose a hazard to us, he quips, you're going to have to pardon my French, but that's complete BS. LeCun's view is that modern LLMs are no more than a fancy simulacrum of intelligence. He acknowledged that they display some features of cat like intelligence, such as persistent memory, reasoning, planning and an understanding of the physical world. However, he noted, we are used to the idea that people are entities that can express themselves or manipulate language, are smart, but that's not true. You can manipulate language and not be smart, and that's basically what LLMs are demonstrating. Basically, Lacune believes that LLMs are fundamentally incapable of reaching AGI, no matter how many GPUs the hyperscalers throw at them, writes the journal. His bet is that research on AIs that work in a fundamentally different way will set us on a path to human level intelligence. These hypothetical future AIs could take many forms, but work being done at Meta's fundamental AI research team to digest video from the real world is among the projects that currently excite Lacune. The idea is to create models that learn in a way that's analogous to how a baby animal does by building a world model from the visual information it takes in. Now, there are many who agree with Lacun's points, there are many who disagree. And then there's the take in the third of the 2018 Turing Award winners, Yoshua Bengio, who said, I hope he is right, but I don't think we should leave it to the competition between companies and the profit motive alone to protect the public and democracy. That is why I think we need governments involved. Anyways, there's nothing particularly new in this article. It's mostly just interesting in the context of the Nobel Prize being awarded last week and that kicking up another round of this debate. Next up, Google is the latest big tech firm to go nuclear to power their AI future. The Hyperscaler has inked a deal with Kairos Power to construct a series of small nuclear reactors. The reactor design uses molten salt cooling technology, which is intended to reduce meltdown risk to zero. Google will help finance and has agreed to purchase energy as an anchor customer. The first reactor will be a 50 megawatt demonstration unit in Tennessee, followed by commercial scale units at 75 megawatt capacity. The deal is for 500 gigawatts in total, with power supply expected to begin between 2030 and 2035. Last year, Kairos was granted the first permit in 50 years to begin construction on a new type of nuclear reactor. They're still awaiting final approval of the design from federal nuclear regulators. Now, unlike Microsoft's deal to restart reactors on Three Mile island, this plan seems more focused on investing in future energy needs in a sustainable way rather than dealing with current AI demand. Google's Senior Director of Energy and Climate Michael Terrell said, we are looking for net new clean power. We are not looking to repurpose existing clean power. If you are looking for one of the big society level conversations that's going to accelerate because of AI, even though it's not exactly about AI, Nuclear has to be at the top of that list. Next up in the world of the constantly shifting talent race between the big AI labs, VP of Genai Research at Microsoft, Sebastian Bubek has jumped ship after spending 10 years at the tech giant. A Microsoft spokesperson said Sebastian has decided to leave Microsoft to further his work toward developing AGI. They added that Microsoft looks forward to continuing their relationship through Bubeck's work with OpenAI. Now this is an interesting one. Bubek had been working on Microsoft's Phi LLMs. These are smaller than traditional Frontier models and are really important for on device AI. Now at this stage, it's unclear whether Bubek will continue this particular work at OpenAI or move on to a new project, but it's still being seen as a pretty big deal. It's also relevant for OpenAI as a lot of what we've been hearing around OpenAI lately is people leaving the company, not people joining the company. On that note, our final story the turnover at OpenAI could be set to continue with reports that former CTO Mira Muradi is recruiting for her new venture. According to the information, Murati has been speaking with OpenAI employees about joining her new venture, though she hasn't disclosed exactly what the venture would be. It's not clear yet whether she's joining an existing company, joining another former OpenAI leader at their new company, or starting something on her own. AI entrepreneur bindu Reddy writes, Mira Muradi, the ex CTO of OpenAI, is raising VC funds and poaching talent from OpenAI to get going on her thing. The more the merrier. Soon we will have over 10 ex OpenAI startups competing with OpenAI. This speaks to how big this market is going to be. That's going to do it for today's AI Daily Brief Headlines edition. Next up, the main episode. Today's episode is brought to you by Venice. The leading AI companies store your entire conversation history and attach it to your identity forever. That's every question you ask, every answer you receive, every image you generate, every thought you share with the machine, it's all being spied on. If you trust all the company's hackers and NSA board members that will ever have access to your AI conversations, then rejoice for you are well served. For the rest of us, Venice is an alternative. Venice is a powerful AI app for text, image and code generation that respects you as a sovereign individual and believes privacy and free speech are not only human rights, but necessary for civilizational advancement. Private, permissionless and uncensored, you can try it for free without an account AI Daily Brief listeners receive a 20% discount on Venice Pro. Visit Venice AI NLW and enter the discount code NLWDAILYBRIEF. That's NLW Daily Brief. All one word. Today's episode is brought to you by Superintelligent. Every single business workflow and function is being remade and reimagined with artificial intelligence. There is a huge challenge, however, of going from the potential of AI to actually capturing that value. And that gap is what superintelligent is dedicated to filling. Superintelligent accelerates AI adoption and engagement to help teams actually use AI to increase productivity and drive business value. An interactive AI use case registry gives your company full visibility into how people are using artificial intelligence right now. Pair that with capabilities building content in the form of tutorials, learning paths and a use case library, and superintelligent helps people inside your company show how they're getting value out of AI while providing resources for people to put that inspiration into action. The next three teams that sign up with 100 or more seats are going to get free embedded consulting. That's a process by which our super intelligent team sits with your organization, figures out the specific use cases that matter most to you, and helps actually ensure support for adoption of those use cases to drive real value. Go to BeSuper AI to learn more about this AI enablement network. And now back to the show. Welcome back to the AI Daily Brief, Today's main episode is about something that's a little bit subtle. At any given time, there's a bit of a give and take in terms of how generally optimistic versus pessimistic people are about not only AI, but technology in general. I think the more that technology has gotten tied up with politics, the more contentious these vibe shifts have been. But I wanted to talk about a subtle shift towards the optimistic side of things that I've seen over the last couple of weeks. Now, I'm not going to go deep into all of the Tesla and SpaceX announcements, but this is certainly part of it. Between an autonomous car and a bus, robots that will be available for sale in a couple of years, and a rocket from space being caught by giant crane chopsticks on its first try, Elon set of companies have certainly contributed a lot to this vibe over the last week or so. Now, for some time, it's been clear that there is a narrative battle around technology. We are in fact around one year from the anniversary of the Techno Optimist Manifesto by Marc Andreessen. He wouldn't have needed to write that if he didn't think that a counternarrative was taking hold. And indeed, it's in two essays today that I wanted to focus on one by Anthropic CEO Dario Amodai and the other by VC Vinod Khosla. Dario dropped his 15,000 word piece machines of Loving Grace last week. Its subtitle How AI Could Transform the World for the Better makes it clear what the agenda of this piece is. Indeed, Dario writes, I think, and talk a lot about the risks of powerful AI. The company I'm the CEO of, Anthropic does a lot of research on how to reduce these risks. Because of this, people sometimes draw the conclusion that I'm a pessimist or doomer who thinks AI will be mostly bad or dangerous. I don't think that at all. In fact, one of my main reasons for focusing on risks is that they're the only thing standing between us and what I see as a fundamentally positive future. I think that most people are underestimating just how radical the upside of AI could be, just as I think most people are underestimating how bad the risks could be. So basically what Dario is doing in this essay is trying to describe, quote, what a world with powerful AI might look like if everything goes right. In other words, and I'll come back to this when we get to some of the critiques of this piece, he is not trying to present a balanced argument Here he's trying to paint a picture of what the best case scenario could be. He breaks his essay into five Biology and Health, Neuroscience and Mind, economic development and Poverty, Peace and governance, Work and meaning. But for our purposes today, I'm going to talk about some of his big takeaways now. First, one thing that's useful is what he means by his terminology of powerful AI. He basically uses powerful AI in substitution for AGI, which he says is a term he doesn't like. He writes, by powerful AI, I have in mind an AI model likely similar to today's LLMs in form, although it might be based on a different architecture with the following in terms of pure intelligence, it is smarter than a Nobel Prize winner across most relevant biology, programming, math, engineering, writing, etc. This means it can prove unsolved mathematical theorems, write extremely good novels, write difficult code bases from scratch, etc. In addition to just being a smart thing you talk to, it has all the interfaces available to a human working virtually, including text, audio, video, mouse and keyboard control, and Internet access. It does not just passively answer questions. Instead, it can be given tasks that take hours, days or weeks to complete, and then goes off and does those tasks autonomously in the way a smart employee would, asking for clarification as necessary. It does not have a physical embodiment, but it can control existing physical tools, robots or laboratory equipment through a computer. He sums it up as a country of geniuses in a data center. Now he also gets into what it's going to take to get there, but since that's really not the point of the piece, I want to focus on these areas of improvement. His first category again is biology and health. He writes, my basic prediction is that AI enabled biology and medicine will allow us to compress the progress that human biologists would have achieved over the next 50 to 100 years into 5 to 10 years. He refers to this as the compressed 21st century. What does that mean? Reliable prevention and treatment of nearly all natural infectious diseases, elimination of most cancer, very effective prevention and effective cures for genetic diseases, prevention of Alzheimer's, improved treatments of most other ailments, biological freedom and doubling the human lifespan. The second category that he digs into is neuroscience and mind. He writes, I expect AI to accelerate neuroscientific progress along four distinct routes, all of which can hopefully work together to cure mental illness and improve function. Those routes are traditional molecular biology, chemistry and genetics, fine grained neural measurement and intervention, advanced computational neuroscience and behavioral interventions. The net result of this Once Again, within that 5 to 10 AI accelerated year period, one most mental illness can probably be cured. Two conditions that are very structural may be more difficult but not impossible 3 effective genetic prevention of mental illness seems possible, 4 everyday problems we don't think of as a clinical disease will also be solved and 5 human baseline experience can be much better. One note on this section, he says, I also suspect that improved mental health will ameliorate a lot of other societal problems, including ones that seem political or economic. Dario's third category is economic development and poverty. Some of the benefits from AI he sees are distribution of health interventions, improved economic growth, food security, mitigating climate change, improving issues with inequality within countries. And he says, overall, I am optimistic about quickly bringing AI's biological advances to people in the developing world. He also says, I am hopeful, though not confident, that AI can also enable unprecedented economic growth rates and allow the developing world to at least surpass where the developed world is now. Number four peace and governance. He writes, suppose everything in the first three sections goes well. Disease, poverty and inequality are significantly reduced and the baseline of human experience is raised substantially. It does not follow that all major causes of human suffering are solved. Humans are still a threat to each other. He basically goes on to point out that although there is a trend of technological improvement in economic development leading to democracy and peace, it is at best a very loose trend with frequent and recent backsliding. Now, this is the most complicated section. In some ways, Dario recognizes that AI is not inherently a democracy promoting technology, even if he sees many of the benefits leading to increased support for democracy as opposed to authoritarianism. However, even in an essay that is theoretically about the upside benefits, this is the one with the most ifs and caveats. Lastly, he talks about work and meaning. Dario writes, even if everything in the preceding four sections goes well, not only do we alleviate disease, poverty, and inequality, but liberal democracy becomes the dominant form of government, and existing liberal democracies become better versions of themselves. At least one important question still remains. It's great that we live in such a technologically advanced world as well as a fair and decent one, but with AIs doing everything, how will humans have meaning? For that matter, how will they survive economically? This, he says, is a more difficult question than the others. And while he gets into a bunch of specifics, I think the most resonant point comes when he writes, I do think in the long run AI will become so broadly effective and so cheap that our current economic setup will no longer make sense and there will be a need for a broader societal conversation about how the economy should be organized. He continues. While that might sound crazy, the fact is that civilization has successfully navigated major economic shifts in the past, for instance hunter gatherer to farming, farming to feudalism to industrialism. He then goes on to speculate around what that shift might look like, but really the point is that we're going to have to have a big conversation about it. Ultimately, he calls all of this a world worth fighting for. He writes, if all of this really does happen over five to 10 years, the defeat of most diseases, the growth in biological and cognitive freedom, the lifting of billions of people out of poverty to share the new technologies, a renaissance of liberal democracy and human rights, I suspect everyone watching it will be surprised by the effect it has on them. I don't mean the experience of personally benefiting from all the new technologies, although that will certainly be amazing. I mean the experience of watching a long held set of ideals materialize in front of us all at once. So what are people's responses? Many in the technology world are represented by Y Combinator partner Jared Freeman, who writes it's hard to overstate just how great the bull case for AI is. Mustafa Suleyman, the CEO of AI at Microsoft, says we must not lose sight of the transformational benefits of AI. Boyan Tungas, formerly of Nvidia, says there have been several other similar essays over the past few months from other top AI voices, but in my opinion, this one is the most thoughtful and most detailed so far. It steers away from many ideological squabbles that have become all too common these days, and provides ample citations to bolster his points and help with further reading and self guided research. Over in the EA, Stephen Pimentel writes, this is the beginning of anthropic's OpenAI arc. I don't mean that in a bad way. Now I can only speculate on what exactly he means by that, but to the extent that one sees OpenAI shifting away from the type of risk focus they had in the past, maybe that's what Steven is referring to. And of course these aren't the only takes. Liron Shapiro, the host of the Doom Debates podcast, says posting a capabilities bull case while sweeping the intractable alignment problem into almost a footnote, he's distracting from the rotten assumption that alignment can happen on the same timeline as capabilities. This is the real Dario. Another Sam Altman figure, another Icarus. Jason Kincaid disagreed a bit, saying, respectfully, I don't think this is an accurate assessment. Anthropic has done more to legitimize the alignment problem than any other lab, and he talks about it in every interview. Nobody will listen to daario if he's Mr. Doomer by painting an optimistic vision. He has more credibility when he says it's time to pump the brakes. That logic makes sense if alignment is tricky but tractable. If it's not, then he's just digging the frame deeper, implying the problem is tractable. 0 acknowledgment that we may be plowing forward to our doom against an intractable problem. And what's sane to do in that case? Andrew Krich writes, I don't think Dario's essay is a distraction, but rather a necessary answer to the question what future are we fighting for? From where I stand, writings like Dario's Machines of Loving Grace are important for helping AI developers to pull together towards a positive vision of the future. Now, as I said, there was another essay that also recently came out by Vinod Khosla, but since this episode is getting long, I will leave that for another time. I'm going to keep coming back to this theme of whether we are seeing a shift to optimism, because I think it's a really interesting one, particularly in light of what I have to assume will be an increase in the regulatory conversations following the election next month. For now, though, that is going to do it for today's AI Daily Brief. Until next time, peace.