5 questions for Austin Carson

How the next wave of technology is upending the global economy and its power structures

Aug 11, 2023 View in browser   By Mohar Chatterjee With help from Derek Robertson Austin Carson. Hello, and welcome back to The Future in 5 Questions. Today, we have Austin Carson — one of the organizers of the White House-endorsed AI red-teaming challenge happening this weekend at the DEF CON hacker convention in Las Vegas, Nevada. Carson is behind the effort to bring in hundreds of college students to participate in this weekend's hacking exercise meant to unearth flaws that the tech industry may have overlooked with the technology. He is the founder and president of SeedAI — a nonprofit that focuses on expanding access to artificial intelligence research and development. Carson previously spent time in D.C. as legislative director for Rep. Michael McCaul (R-TX), executive director for the Technology Freedom Institute, and led NVIDIA’s government affairs operation.   Read on to hear Carson’s thoughts on using AI to speed up scientific discovery, the very human reaction to the technology’s limits, and the federal government’s problems keeping up with  tech progress. Responses have been edited for length and clarity. Listen to the rest of the interview in today’s POLITICO Tech podcast.  What’s one underrated big idea? I can't say if it's underrated, or simply under-discussed, but the application of artificial intelligence to the scientific exploratory process to dramatically transform our research. It's easy to lose track of the incredible rate of AI research itself. But in science, many foundational progress points have been through artificial intelligence. One good example: Anima Anandkumar at Caltech used AI to solve partial differential equations far more efficiently. That's effectively a time machine for anything involving fluid dynamics — which is a whole lot of things. So we’re getting into this recursive effect where AI gets better at directed and then autonomous scientific discovery to some extent. That requires responsibility, that requires oversight. But our ability to transform the fabric of things is moving rapidly. What’s a technology you think is overhyped?  Humanoid robots. We spent a lot of time talking about robots that look and move like us. But if you think about a lot of applications — even for general-purpose things, it wouldn't necessarily make sense to recreate our form and process. So, the fixation on AI as the “Terminator” is both overrated and not necessarily useful. What book most shaped your conception of the future? “The Three-Body Problem” and a splash of “Foundation” on the side. The “Foundation” series by Isaac Asimov gives a great idea of sprawling progress combined with human interaction. They made a show, but it's worth reading the series because it's from the 1950s and there are so many interesting things still tied to it. “The Three-Body Problem” was just an incredible and specific thought experiment of what it could be like for us to encounter alien life and dramatically more advanced physics improvements. It’s a very interesting take on what it’s like to be in China and Chinese culture. But it has three books and you’ve got to read all three. Even if the first one's not your favorite, you’ve got to read the next two. What could government be doing regarding tech that it isn’t? A key problem of what the government isn't doing about technology is the fact that nobody right now knows what the hell to do about many technological advancements. They're moving so rapidly — it's hard to be nimble. By that, I mean the federal government is designed — purposefully so — to not move in week-to-week policy shifts, just because a new piece of AI research came out. Part of what the government isn't doing is exploring all possible avenues to bring in more smart people. Great policy support right now is done by fellows who come and participate through programs like TechCongress or Horizon or PIF — the Presidential Innovation Fellowships. That's still just a headcount thing. So how do we find ways to meet the methodical, deliberate process of the federal government? That’s a creativity exercise, and that's on all of us. What has surprised you most this year? I was not surprised most by the sudden public explosion of interest in artificial intelligence. It was like when you watch a movie with a bad twist. What has been very surprising to me was the very human reaction to the news by otherwise intelligent people. Everybody kind of suddenly moved to, “Oh, it can't pass the MCAT? It's not better than me. Oh, wait, it can't pass the bar? And give me relationship advice? And tell me how to fix my VCR?” People are weirdly insecure about it. But I'm on panels and the more technical a person is, the more philosophical they have gotten. We've gotten to the point where we kind of understand the tech, but now — we have to think about a lot of things involving ourselves.   A NEW PODCAST FROM POLITICO: Our new POLITICO Tech podcast is your daily download on the disruption that technology is bringing to politics and policy around the world. From AI and the metaverse to disinformation and cybersecurity, POLITICO Tech explores how today’s technology is shaping our world — and driving the policy decisions, innovations and industries that will matter tomorrow. SUBSCRIBE AND START LISTENING TODAY.     keeping the ai hype honest in cybersecurity LAS VEGAS — After a decade in which artificial intelligence was supposed to revolutionize cybersecurity, the technology could finally be poised to deliver on its promises, according to Meta senior AI scientist Joshua Saxe. But to get there, the industry needs to be realistic about what AI models are capable of. Back in 2012, ex-Google researcher and machine learning pioneer Geoffrey Hinton and his colleagues published ground-breaking work that used artificial intelligence to sort high-resolution images into a thousand categories (albeit with high error rates, by today’s standards). Hinton predicted the technology would someday render professions like radiology obsolete. That prediction has yet to come to pass, and Hinton's remarks became emblematic of how the hype around technology can blow past its real-world applications, Saxe said. That hype slowed down the process of getting to realistic uses for AI in cybersecurity, affecting what ideas went from prototype to production. “In retrospect, while we achieved a lot, I wish that there had been more open, honest discussion about the apparent limitations of this technology,” Saxe said, speaking to a packed room at DEF CON. Saxe presented his case that the groundswell of technical advances is finally catching up with the decade-old hype. AI models today are more capable of sophisticated data classification, reverse engineering how malware works, and detecting phishing attempts. And large language models trained on vast reams of data can process information far more dynamically than their predecessors. Another factor that will distinguish the current moment from a decade ago is the explosion in financial investment the field is receiving, Saxe said. That flow of capital will further accelerate the pace of AI research. Hurdles still remain, though, he said. Since training these models from scratch is a formidable undertaking, the latest generation of LLMs are frozen in time based on when they were trained, and their reasoning process can be unreliable. For those in the cybersecurity world, those hurdles are a reminder to stay realistic about AI’s potential applications. "I don't think we could have avoided fumbling around in the dark to find applications — but I think if we had been even more ruthlessly honest with ourselves, we could have probably been more efficient in our search,” Saxe said. — Mohar Chatterjee keep on truckin' Self-driving cars are so back. … On the streets of San Francisco, that is, as California’s Public Utilities Commission voted yesterday to authorize robotaxi service at all hours of the day and night, overriding the protests of city and fire officials who claim they’re a safety hazard. And labor groups aren’t happy either. As POLITICO’s Jeremy B. White reports, the conflict over autonomous vehicles isn’t just over safety but a microcosm of a larger economic fight in California between labor and the state’s powerful tech sector. “Not only do our members share the roads with Robotaxis, but the technology can be deployed to map, replicate, and then eliminate traditional package delivery,” Peter Finn of the California Labor Federation, a group representing California’s labor unions, said in a text message. “This is an existential threat to hundreds of thousands of good middle-class jobs across California communities.” There’s a statehouse fight brewing over automated trucking, as well, Jeremy writes. In addition to having attempted to pressure the state to block the San Francisco robotaxi authorization, they’re pushing a bill that would require automated trucks to have human drivers on board. — Derek Robertson Tweet of the Day THE FUTURE IN 5 LINKS Take a closer look at the VR gizmos on display at SIGGRAPH this week. A Wired writer argues we need a new Turing test for today’s AI. Does the AI boom have anything in common with the dot-com bubble? AI bots with human names pose thorny semantic problems. Powerful new software is revolutionizing mass production. Stay in touch with the whole team: Ben Schreckinger (bschreckinger@politico.com); Derek Robertson (drobertson@politico.com); Mohar Chatterjee (mchatterjee@politico.com); and Steve Heuser (sheuser@politico.com). Follow us @DigitalFuture on Twitter. If you’ve had this newsletter forwarded to you, you can sign up and read our mission statement at the links provided.   HITTING YOUR INBOX AUGUST 14—CALIFORNIA CLIMATE: Climate change isn’t just about the weather. It's also about how we do business and create new policies, especially in California. So we have something cool for you: A brand-new California Climate newsletter. It's not just climate or science chat, it's your daily cheat sheet to understanding how the legislative landscape around climate change is shaking up industries across the Golden State. Cut through the jargon and get the latest developments in California as lawmakers and industry leaders adapt to the changing climate. Subscribe now to California Climate to keep up with the changes.       Follow us on Twitter Ben Schreckinger @SchreckReports Derek Robertson @afternoondelete Steve Heuser @sfheuser   Follow us

To change your alert settings, please log in at https://www.politico.com/_login?base=https%3A%2F%2Fwww.politico.com/settings This email was sent to rouf@idiot.cloudns.cc by: POLITICO, LLC 1000 Wilson Blvd. Arlington, VA, 22209, USA Please click here and follow the steps to unsubscribe.