Theo Jaffee on Becoming AGI-Pilled

Theo Jaffee, a 21-year-old University of Florida dropout, co-founded Monitoring The Situation (MTS), an Andreessen Horowitz-backed media company that streams interviews with AI and tech figures on X, after becoming convinced by ChatGPT that AI was the future. Jaffee left college in 2023, moved to San Francisco, and built a following through his podcast, interviewing prominent thinkers like Zvi Mowshowitz and Robin Hanson before launching MTS.

Present at the Creation Postscript No. 22 | Theo Jaffee on documenting the Singularity POSTSCRIPT Two years ago, the chief-of-staff to Anthropic’s CEO declared https://www.palladiummag.com/2024/05/17/my-last-five-years-of-work/ , “These next five years might be the last few years that I work.” At the time, she spoke for a world that knows far more about the future than the average American. But even before the release of ChatGPT, Theo Jaffee knew that the future was happening in the West. The 21-year-old University of Florida drop-out found his computer science classes in Gainesville pointless; they operated under old assumptions about how to build a career. Come summertime, he left for San Francisco, drawing on connections with former Curtis Yarvin intern Stevie Miller, attending the Manifest conference in Berkeley, and building out his Twitter following. Now Jaffee is the co-founder of Monitoring The Situation MTS , an Andreesen-Horowitz backed media company that streams interviews of central players in AI and technology all day on X formerly Twitter . Jaffee is San Francisco’s go-to reporter and AGI discussant, tirelessly cataloguing, describing, and perhaps influencing the future of our world. A Spanish ruler declared long ago, “Had I been present at the Creation, I would have given some useful hints for the better ordering of the universe.” But if you cannot be present at the creation — well, at least you can monitor the situation. Below, I talk to Jaffee about becoming “AGI-pilled” and what we should do to prepare for a superintelligence we’re all underestimating. Our conversation has been edited for length and clarity. New Critic paid subscribers get access to Postscript, our interview series, Contra, our criticism section, and exclusive New Critic parties for only $30 a year. ELAN How did you get to the center of the tech world? THEO I went to the University of Florida in fall 2022, and that was when ChatGPT came out. When I saw ChatGPT for the first time, I realized, “Holy shit, this is going to be the future.” I spent the next few months scrolling Twitter constantly, meeting mutuals virtually. I was in Gainesville, Florida, so there was no one to meet in real life; everyone’s in SF or New York. I was reading LessWrong, discovering this whole ecosystem of people, including Curtis Yarvin — I read Yarvin a lot at that time. This was my freshman year of college. I didn’t have an internship for that summer, so I decided — kind of like one of these people I knew from Twitter, Dwarkesh Patel — I would start a podcast where I’d find people I liked on Twitter and interview them. This was summer of 2023, three years ago now. I started interviewing a few people. My first guest was Greg Fodor. Some of the more famous people I interviewed were Zvi Mowshowitz, Robin Hanson, Bryan Caplan, and so on. I did maybe 20 episodes over the two years of the podcast — I definitely should have done more. In August 2023, I visited SF for the first time. I didn’t meet any mutuals then, but I discovered the city and fell in love with the place. I was like, “Wow. It’s beautiful, it’s full of technology and tech people — I definitely need to be here after college.” I don’t love the city as much anymore, but I still love the people. Throughout my sophomore year — fall 2023 through spring 2024 — I continued being really interested in AI and AGI. This was the GPT-4 Turbo era, and then Claude 3 Opus came out around that time, too. I continued reading Yarvin and other right-wing people like Richard Hanania. That summer — May 2024 — I again didn’t have an internship. I had a few thousand dollars, and I figured if I had nothing else to do, I might as well go to SF full-time and do my podcast, hoping something would fall into place. My Airbnb for the summer turned out to be a fake listing, so I had to stay in a hostel with shared rooms for almost three weeks until I found a more permanent place, which was the Minerva building at 16 Turk Street in the Tenderloin — it was ancient, decrepit, had no amenities, and I shared a dorm-style room with one other guy. It was $950 a month. That was summer 2024. When I got to SF, I had about a thousand Twitter followers. I posted some good tweets in the first couple weeks that doubled me to about 2,000, and I started meeting people in real life, partly through Manifold Markets’s Taco Tuesdays, hosted by Manifold. I got to know Stephen Grugett and Austin Chen and, to a lesser extent, James Grugett. I met a lot of Twitter people, tech people, rationalists, in real life. In May, a couple weeks after I arrived, I went to a Scott Alexander meetup at Lighthaven in Berkeley. It was my first time there. I fell in love with the place. There were so many people there I knew from the internet, and the venue itself is beautiful. That weekend was Manifest. I had discounted student tickets. It was the greatest time of my life. It was incredible to be around all these people who lived and breathed the stuff I’d been reading and writing about on Twitter. By that point, I’d been on Twitter for almost two years and had been posting for about a year. I should mention that I heavily credit Stevie Miller, above just about anyone, for helping me get into this. We’ve known each other since we were about eight. We went to the same elementary school in Florida for a couple of years. He moved to D.C. but kept visiting because his grandmother lived in Florida. He introduced me first to Neoreaction, then to Curtis Yarvin, then to the SF rationalists. I found LessWrong independently, but he showed me it was a whole movement — told me about Manifold Taco Tuesday, told me about Manifest. I owe an immense debt to Stevie. During summer 2024, I applied for and got an Emergent Ventures grant from Tyler Cowen’s Mercatus Center. That’s what let me survive in SF the rest of the summer. I came back to school after that, constantly thinking about going back to SF. During my junior year, I was still very interested in AI, still making a few podcast episodes — though not as many as I should have — and still wondering how I’d get back to SF. I spent that winter break kind of bumming off my friend’s couch in Berkeley and hanging out at Lighthaven during the day, and I think I visited again over spring break. Once again, I had no internship lined up for that summer. For a while, I was so pessimistic about AI’s impact on jobs in the near term that I thought, “What’s the point of even trying a normal software engineering career?” It turns out I was wrong but not super wrong. What I’m doing now is probably better for me in the long run than if I’d just gotten a job at Google or somewhere as a normal software engineer. I also have to credit Erik Torenberg immensely. Back in summer 2024, Erik — who, at the time, ran Turpentine, his podcast network — was a big fan of my podcast. He’d respond to episodes, saying “Wow, great episode.” My podcast had very few viewers — my most-viewed episode ever on YouTube was about 3,000 views — so for Erik to be that tapped in was impressive. In February 2025, he got hired as a general partner at Andreessen Horowitz a16z . In April, a couple months later, he invited me to intern at a16z — of course I said yes. This completely saved me, since I had nothing lined up for that summer. After a lot of back-and-forth with HR — it took about six weeks from his first message to my start date, since Erik didn’t know if he was allowed to hire a summer intern, or if he had to hire full-time. He hired me as an intern for the summer on the New Media team. I was mostly helping them with research — one of the media team’s things was the a16z Podcast. When we had guests on — people like Sam Altman — I helped write research briefs and questions. As time went on, they gave me a return offer to go full-time in the fall, in August, right before my 21st birthday. I joined full-time on the New Media team doing that research, plus special projects, like helping a lot with the New Media fellowship. ELAN How have your views of AI evolved over time? THEO As a young teenager in 2017-2018, I was aware of the idea of superintelligence. I’d watched the CGP Grey video “Humans Need Not Apply,” read the Wait But Why series on superintelligence, and I knew that probably some decades out, but within my lifetime, superintelligence would become a thing. Around 2018-2019, I played with some AI tools — Talk to Transformer, AI Dungeon based on GPT-2 . I thought they were interesting toys that didn’t totally grab my attention. I played with them a bit more in 2020-2021, was aware of GPT-3’s launch, had an OpenAI API account, played with that a decent amount — tried using it for copywriting for my dad’s business, for college essays, for homework in high school in 2021-2022. I thought, “It’s basically a toy. It hallucinates, makes mistakes.” I didn’t really think about what scaling would do. People don’t remember this, but in mid-2022 Stable Diffusion was released, and I thought, “Okay, AI can make images. This is worth keeping an eye on.” In November 2022, Meta released Galactica — nobody remembers this, but this was ChatGPT right before ChatGPT. I thought it was super sick. They ended up pulling it because it hallucinated citations, which was a problem because it was designed as a scientific research tool. Then ChatGPT came out. Different people have different moments that make them an AGI-pilled person, and for me, that was it. I realized, “Holy shit, this is real. The computer is talking like a human, taking basically anything I throw at it, and writing genuinely good responses.” At the time, it couldn’t code or do economically useful white-collar tasks, but you could have a conversation with it, ask it about history and science and get a perfectly good answer as long as it was qualitative. I realized, “This is literally going to change everything.” It was the first step toward the superintelligence future I’d read about years earlier. I remember calling my dad right after ChatGPT came out and saying “Holy shit, this changes everything.” He prompted it with something like, “Write a sermon from a rabbi about such and such topic,” and it “pilled” him, too.