The AI Daily Brief: Artificial Intelligence News and Analysis - AI's 100 Most Influential (According to TIME)

Episode Date: September 9, 2023

Today NLW looks at TIME's first ever list of the 100 most influential people in AI, including some of the most notable snubs. Before that on the Brief, two senators prepare to introduce comprehensive AI legislation. Today's Sponsor: Supermanage - AI for 1-on-1's - https://supermanage.ai/breakdown ABOUT THE AI BREAKDOWN The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://theaibreakdown.beehiiv.com/subscribe Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown Join the community: bit.ly/aibreakdown Learn more: http://breakdown.network/

Today on the AI breakdown, we're looking at the Time AI 100 list. Before that on the brief, two senators are getting ready to propose comprehensive AI legislation in the U.S. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Go to Breakdown.network for more information about our YouTube, our newsletter, and our Discord. Welcome back to the AI breakdown brief, all the AI headline news you need in around five minutes. Today we begin on the policy side of the House, where obviously a huge, huge push in the United States right now is AI legislation. This is happening in really two different

ways. The first is that there are many bills being proposed that deal with some smaller aspect of artificial intelligence or fallouts from the industry. We talked earlier this week about the attorneys general of 50 states in the U.S., sending an open letter to Congress asking them to increase protections around AI-created child porn, for example. But then the other side, and the much more significant legislative efforts, are looking at comprehensive regulatory frameworks. These are frameworks that would address questions like what the requirements are for advanced models, what sort of disclosures there need to be around training data, you know, the big thorny questions that will really shape the industry. According to reporting from the New York Times, Democratic Senator Richard Blumenthal and Republican Senator Josh Hawley are on the verge of introducing a comprehensive framework for AI regulation. Now, these are the leaders of the Senate Judiciary Subcommittee for Privacy Technology and the Law, and they're the ones who have been most active holding hearings around artificial intelligence.

They hosted Sam Altman in May and are having another hearing on Tuesday, which will include the chief scientist from Nvidia, Brad Smith, Microsoft's president, and apparently at that hearing, they plan to discuss a number of the parts of their proposed legislation. The New York Times discusses four aspects of what that framework will likely include. By far the biggest and the one that's getting the most attention is the creation of an independent federal office to oversee artificial intelligence. Now, this is something that people like Sam Altman have expressed some interest in, while others, including IBM and Google, have pushed strenuously against the creation of a new office, saying that other agencies have everything they need to regulate the industry right now.

Other topics that will be addressed in the legislation include licensing requirements for new AI models, new policies around the liability that companies face around potential privacy and civil rights violations, and new requirements around data transparency and safety. Now, next week is shaping up to be a big one for AI in Washington, because in addition to that Tuesday hearing, on Wednesday, Senate leader Chuck Schumer is holding a closed-door meeting on AI regulations that will feature the participation of a number of leaders in the industry, including Elon Musk, Microsoft CEO Satya Nadella, OpenAI Sam Altman will be back, Zuckerberg will be there, and many others as well. Now, speaking of governmental efforts around AI regulations and safety, earlier this summer, the UK announced their foundation model task force and named entrepreneur Ian Hogarth as chair.

Yesterday, Ian tweeted, 11 weeks ago, I agreed to chair the UK's efforts to accelerate state capacity and AI safety, measuring and mitigating the risks of frontier models so we can safely capture their opportunities. Here is our first progress report. First, they say they've established an expert advisory board that includes notables from the world of AI and AI safety, including Turing Award winner, Joshua Benjio, AI alignment researcher, and a number of others. They've recruited a group of what they call expert AI researchers, who will be focused on working on safety issues for the UK government. They've partnered with a number of leading technical organizations, and they are working on, quote, building the technical foundations for AI research inside the government. Basically, this means getting these researchers access to the compute to actually keep up with the private sector, which is going to be no mean feat. Now, in just a couple months, the UK will have its big AI safety summit, so I'm sure we'll get another update on what the Foundation Model Task Force has been up to at that point. Moving over to the private sector, Microsoft has, in collaboration with Page,

announced what they're calling the world's largest image-based artificial intelligence model for digital pathology and oncology. This was an LLM that was trained using over a billion images from half a million pathology slides across multiple cancer types. Now they're working with Microsoft to create a new AI model that they say as orders of magnitude larger than any other image-based AI model existing today. Their goal is to, quote, capture the subtle complexities of cancer and serve as the cornerstone for the next generation of clinical applications and computational biomarkers that push the boundaries of oncology and pathology. So basically

Page, which is a spin-off from the Memorial Sloan Kettering Cancer Center, is building a new AI tool for doctors that helps fight cancer. Microsoft is a partner not only for the development of the model, but also for deploying it via Microsoft Azure. Microsoft was also in the news, when it followed Adobe in saying that enterprise users of its AI tools, in this case, co-pilot, will be protected from legal indemnity, or at least that Microsoft will pay if they get sued for copyright infringement. It seems that there is a sense in the corporate world that AI generated anything is likely to be subject to lawsuits. And so to get out ahead and get a foothold in the industry, it's worth it to these companies to mitigate that risk for their enterprise buyers,

if only in the short term. In the realm of LLM Competition, Anthropics Claude has now introduced a pro model costing $20 per month. The verge says that the main draws that you get five times the usage with Claude Pro as compared to the free tier. And this is obviously meant to help Claude compete with ChatGPT Plus, but of course the real question when it comes to competition is going to be less about price and availability and more about differentiated features. Initially, many were attracted to Claude 2's bigger context window, which is at 100K tokens, for use cases like larger document analysis. And finally, speaking of AI startups, we have a new freshly minted AI unicorn. Mbue, which was formerly known as Generally Intelligent, has raised a $200 million series B that values the company at over a billion dollars.

Now, Embu launched last October, and its goal is to create AI agents that can actually do things in the real world. In a blog post announcing the funding they wrote, our goal remains the same, to build practical AI agents that can accomplish larger goals and safely work for us in the real world. Now, one of the interesting things about Mbue is their thesis on why AI agents aren't particularly effective so far. In their announcement post, they wrote, we believe reasoning is the primary blocker to effective AI agents. Robust reasoning is necessary for effective action. It involves the ability to deal with uncertainty to know when to change our approach, to ask questions and gather new information, to play out scenarios and make decisions, to make and discard hypotheses, and generally to deal with the complicated, hard-to-predict nature of the real world. The company is really focused on coding use cases,

not just because coding is a valuable application of AI, but because they argue, coding is a vehicle for improving reasoning. The company wrote, an agent that writes an SQL query to pull information out of a table is much more likely to satisfy a user request than an agent that tries to assemble that same information without using any code. Moreover, training on code helps models learn to reason better. Training without code seems to result in models that reason poorly. Now, this is sort of validated by the lived experience of people who have used chatyBT with code interpreter versus without, who find that the addition of the ability to code actually opens up an entirely new set of use cases. In any case, the $200 million

investment shows that there is still a, Bucco Bucco's in AI Venture funding generally, and B, that AI agents are still very top of the heap in terms of things that people are interested in. However, that is going to do it for this AI breakdown brief. Thanks for listening or watching as always, and I'll be back soon with the main AI breakdown. Before we get into the main AI breakdown, I want to tell you about today's sponsor, Supermanage. If you work in a professional setting, you probably have some version of a one-on-one meeting, either with the people that work for you or the people that you work with. Unfortunately, all too often those one-on-one meetings become glorified catch-up calls.

Don't you wish you could jump right to the stuff that really matters? That's where Supermanage comes in. Supermanage AI magically distills your team's public Slack channels into a real-time brief on any employee, any time. Catch up on contributions, work in progress, challenges their facing, sentiment, everything you need to show up ready for a truly meaningful conversation. And it's completely free. Visit supermanage.aI.4.3 breakdown today to start

making the most of your one-on-ones. And thanks again to Supermanage for sponsoring the AI breakdown. Welcome back to the AI breakdown. Today we are doing a deep dive on Times AI 100. This is a list that Time magazine has curated of who they call the 100 most influential people in artificial intelligence. So what we're going to do today is look a little bit about how they categorize things, the way that they describe their methodology, any notable entrance to the list, any notable exclusions from the list, and overall what the response has been. If nothing else lists are a great thing to debate, so this should be some good fodder for that. Now, let's talk first about what Time says about how they chose this list. They wrote,

Times' most knowledgeable editors and reporters spent months fielding recommendations from dozens of sources to put together hundreds of nominations that we whittled down to the group you see today, said their executive editor, we wanted to highlight the industry leaders at the forefront of the AI boom, individuals outside these companies who are grappling with profound ethical questions around the uses of AI and innovators around the world who are trying to use AI to address social challenges. This group of 100 individuals time continues is in many ways a map of the relationships and power centers driving the development of AI. They are rivals in regulators, scientists and artists, advocates, and executives. The competing and cooperating humans whose insights,

desires, and flaws will shape the direction of an increasingly influential technology. So two takes right up front from me on this. One, you have to remember that lists like this are done because they are an extremely compelling form of content. This is not some mechanical, scientific ranking of the exact 100 most influential people. This list is instead meant to be subjective, to have opinions. There is a sense among many that I've seen that if this list were a pure scientific exercise, it would basically be 100 researchers and developers. And that might be true, but that's clearly not what time is talking about. They're talking about not only artificial intelligence, but artificial intelligence in its societal context. And what's more, it's clear that they are also trying to

show a cross-section of the domains of human experience that AI will ultimately interact with and shape. The inclusion of the, quote, innovators around the world who are trying to use AI to address social challenges makes that goal of this list playing. So let's dig into the list itself and see how they broke it apart. The first category are leaders. And this, frankly, is the section of the list that probably has the greatest number of people that you would simply expect to be here. You've got Dario and Daniela Amodi, the CEO and President of Anthropic. You've got Sam Altman, the CEO of OpenAI, Demis Havas, the CEO and co-founder of Google Deep Mind. One thing that I was glad to see was a recognition of the importance of open source in the conversation

with the recognition of Clement DeLang, the CEO and co-founder of Hugging Face. In their interview with Clement, he gives a strong, simple version of the argument that AI power concentrated in a small number of hands is more dangerous than Open AI that is dispersed and available to many. He says, if you look at society, the biggest risk is actually to have power and understanding concentrated in the hands of a few, especially for a time. technology like AI, and especially if these organizations are not designed to cater for the public good. Companies might have good intentions, but they are private profit-seeking organizations

by nature. If you look at long-term healthy development of the technology, we believe that more democratization creates more counterpowers and fewer risks, because it empowers and enables regulation. Regulators can't regulate something that they don't understand that they don't have transparency into. Another person that you would expect to show up and who indeed does is Mark Andresen, who serves here in many ways as representative of the accelerationist point of You can tell time has a bit of a skeptical attitude towards Mark here. They write, unsurprisingly, Andresen is selling a vision of the future in which he is heavily invested. They also point out, of course Andresen has also been wrong in the past. Andresen Horowitz was one of the main drivers

of a crypto bubble that popped last year. Bitcoin hasn't replaced cash. And Coinbase, a crypto exchange that Andreson has heavily invested in, hasn't replaced banking, or at least not yet. Another investor that I thought was a notable inclusion was Daniel Gross. Daniel and fellow investor Nat Friedman this year announced Andromeda, which time calls, quote, a mountain of cutting-edge chips weighing 7,255 pounds and costing around $100 million. Continuing time rights, the chips are wired together to form a giant compute cluster, to which Gross and Friedman trade access in exchange for equity in AI startups they judge to be promising, a move that other venture capitalists are considering aping. Basically, I think that

this inclusion recognizes just how important actual access to compute is in the development of AI right now. And again, there are a lot more expected folks here. Kevin Scott, the CTO of Microsoft, Jensen Huang, the CEO of Nvidia, etc. etc. Et cetera. Now, what about the more nebulous category of innovators? This is certainly the category where more debate has come up, I think. This category includes numerous artists and musicians, including Grimes.

Now, I saw a bunch of people basically saying, go home time, you're drunk, when they put Grimes on the Time 100 AI list. But I think this one kind of makes sense. Again, this is in the context of trying to help people understand the breadth of how this technology is starting to interact with and influence culture, as well as just the industry. And as the entire music space was freaking out about Heart on My Sleeve, Grimes ran in the other direction and actually came out and said that she would collaborate with anyone who wanted to train AI on her vocals and music and share royalties with them if she thought the songs were good enough.

She even ended up creating an AI model of herself that people could use to make that process more easy. One inclusion that I think represents this idea of people using AI in the context of big problematic social issues is Nathaniel Manning, COO, and co-founder of Kettle, which is trying to use AI to create what time calls a more nimble insurance market. Basically, Manning and Kettle argue that the way that insurance works using models based on historical data doesn't make sense in the context of climate change. Time writes, in addition to historical data, Kettle is using satellite imagery, weather data, and machine learning technology to form what it says is a more accurate picture of the wildfire risks facing California homeowners. This allows the company to offer affordable insurance to homeowners

in the state deemed risky by other insurers methods, but relatively safe, to Kettle. Time writes, the hope is that the method will create an insurance market that actually prices in climate risk, incentivizing people to move to safer areas. So again, the point is that when it comes to the next wave of AI models, it's very unlikely that what Nathaniel Manning thinks is going to shape what the big tech companies do and what you're using in a year or two. But here he represents the way in which an entire group of people are bringing AI into the climate conversation and more broadly other issues of social importance. The shapers section is where policy lives. It features folks like Ian Hogarth, the chair of the UK's Foundation Model Task Force, who we actually just talked about in the brief today. It includes the heads of policy at companies like Anthropic and Open AI, and it includes just one sitting U.S. member of Congress, in this case, Anna Eschew. Now, given that senators Richard Blumenthal and Josh Hawley are about to introduce comprehensive legislation, we're obviously going to have many more voices in Congress and the Senate when it comes to AI. But again, in many ways, this is a stand-in for all of those voices, right? And the fact that this has now become an incredibly important issue in Congress and the Senate.

The Shaper's section is also where folks who are involved in AI safety live. Tristan Harris, who's the founder of the Center for Humane Technology, and of course Eliezer Yudkowski, whose essay in time earlier this year was another pivotal moment in the rise of AI safety discourse going mainstream. Now, obviously, things get a little blurry between these categories, given that Eliezer lives over in the Shaper's section, but Jeffrey Hinton, who I think is most responsible this year, at least, for the rise of the AI safety discourse going mainstream, lives over in the thinker section, representing a very deep.

different set of thinking on the AI safety conversation is Jan Lacoon, who is also a Turing award winner, but who has called existential feels around AI as, quote, an apocalyptic cult. Now, one of the best questions that Time asks, Lacoon, was why do you think your viewpoint on this diverges so drastically from the other godfathers of deep learning Jeffrey Hinton and Joshua Benjillo, both of whom shared the Turing award with him in 2018? Jan said, Jeff believes LLMs are smarter than I believe they are, and he's a little bit more optimistic than I am about how they might get us to human-level AI. So he realized all of a sudden, quote, we need to worry about super-intelligent machines and if you have a more intelligent entity, it is going to want to take over the world.

And I think that's just wrong. There's no correlation between being intelligent and wanting to take over. Even within the human species, it's not the most intelligent among us who want to be the leaders. In fact, it's quite the opposite mostly. The desire to dominate is really attached to species that are hierarchically organized and social. It's really a consequence of human nature and the fact that evolution built us this way. But orangantangs, for instance, are not a social species and don't have any desire to dominate anybody. They don't need to. So we can be smart like orangutans without having any desire to dominate. Now, one of the things that I think is notable about this aside from the unknowability of it is the extent to which we're debating based on questions of human nature and

evolutionary biology as much as current technology. More open AI folks show up in this category as well, including the co-leads of their superalignment team, Jan Lakey and co-founder Ilya Sutskever. And ultimately there is only one sort of media creator-ish inclusion, which is Arvin Narayanan. Now, these are a professor of computer science and one of his PhD candidates at Princeton, and they write a substack called AI Snake Oil. They're basically warning against the hyperbole that is so profligate in AI-related media, and in so doing, I think, offer something valuable. It's something that I'm subscribed to for sure.

Still, I hope that by next year's time, AI 100, maybe there will be more space for media creators who are not just outright skeptical, but who are still helping people navigate what is an incredibly important shift. Now, when it comes to snubs, there is no doubt that the biggest disagreement was another open AIer, Andre Carpathy. Pratik to Sai tweets, no Carpathy? Blasphemy. And almost 2,000 people liked that tweet. Former pharma exec who's now building an AI startup in the health field, Martin Schrelli, writes, useless list. Now, ultimately, again, I caveat this and think you should too with understanding that this is not just an intellectual product, but this is a content product. This is a list that's meant to introduce a very normy audience that reads time magazine.

magazine to a wide array of people in the AI space who are shaping the discussions, and frankly, I think in that they were successful. What's more? Debate is very good for Time Magazine as it just drives more attention and clicks, so by all means keep debating, and if nothing else recognize that the fact that this cover exists and that this list now exists, suggests just how fast this technology has entered into the public consciousness. If you want to join the debate more directly, come on over to the AI Breakdown Discord. You can find it at bit.ly slash AI breakdown and you can share with our community all of the people that you think should be on this list, who you definitely think shouldn't, and how you would of course do it better. For now,

though, I am going to wrap there. I appreciate you guys listening as always. And until next time, peace.