Faster, Please! · Economics & Policy
TIER 4 Sun, 28 Dec 2025 21:02:49 +0000
As 2025 ends, AI's superintelligence moment still lies ahead. But how far? ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ | | ---|---|--- | | | Forwarded this email? Subscribe here for more --- # ✨ Waiting for AGI, again ### As 2025 ends, AI's superintelligence moment still lies ahead. But how far? Dec 28 --- --- | | | --- | | --- | | --- | | --- | | READ IN APP --- | | ---|---|--- _🎄 The holiday season is here, which is especially good news for free**Faster, Please** free subscribers like yourself._ _Between now and New Year 's,**you can get a 50% discount on a paid subscription**!_ _Just hit the Big Blue Button below for more info. My guess is that 2026 is going to be an amazing year on the pro-progress front, and I can 't wait to cover and analyze it for you folks!_ _I appreciate your support!_ _Jim_ Get 50% off for 1 year * * * **My fellow pro-growth/progress/abundance Up Wingers in the USA and around the world:** Steel yourself! Buckle up! Prepare for impact! The end of 2025 means we're only a year away from this world: Advances in artificial intelligence turn science-fiction scenarios into shocking news headlines. AI systems begin improving themselves, automating much of white-collar work, and reshaping advanced economies like America's almost overnight. Governments scramble to keep up as fears grow that leading-edge systems don't fully align with human goals. Early optimism about the arrival of artificial general intelligence (Warp-speed growth! Miracle cures! Fusion reactors,_tout de suite_!) gives way to public backlash and panic (Nationalize the AI labs!). In retrospect, the AI data-center NIMBY movement in 2025 was a Down Wing harbinger. All that -- at least if the most likely AGI timeline sketched in the much-cited AI 2027 forecast from the AI Futures Project turns out to be right. But what if that modal prediction is actually a bit too pessimistic about the pace of progress? Some holiday season scrolling will find lots of aggressive takes by "Singularity soon" enthusiasts. A few examples: * "I predict that AGI and ASI will be achieved by the end of 2026." * "In 2026, the future stops arriving slowly and starts arriving all at once." * "2026 will be the year AGI is widely reported. Buckle up, because the next step is something much smarter than all of us. What a time to be alive!" And there's also this headline: "Elon Musk Predicts AGI by 2026 (He Predicted AGI by 2025 Last Year)." I think that bit of skepticism is appropriate, and not just because of Musk's track record of ambitious forecasting. Look, I'm not a scientist or technologist. I don't feel qualified to independently judge the AI frontier. So instead I mostly focus on how AI advances are manifesting (or not) in the economy by looking at what companies are doing, what CEOs are saying (and not just the biggest names in AI), and what economic statistics are showing, or at least suggesting. So business bottom lines over academic benchmarks. As far as gauging the state of the frontier, I emphasize betting and prediction markets. Here's a representative selection of what those platforms are saying about the state of play right now: * "By 2028, will there be a visible break in trend line on US GDP, GDP per capita, unemployment, or productivity, which most economists attribute directly to the effects of AI?" 35 percent. (Manifold Markets) * "When will OpenAI achieve AGI?" 11 percent before 2027, 24 percent before 2028, 41 percent before 2030. (Kalshi) * "Conditional on no existential catastrophe, will there be a superintelligence by 2030?" 22 percent. (Manifold Markets) * "OpenAI announces it has achieved AGI before 2027?" 13 percent. (Polymarket) * "When will the first weakly general AI system be devised, tested, and publicly announced?" December 2027. (Metaculus) * "When will the first general AI system be devised, tested, and publicly announced?" July 2033. (Metaculus) But what says the largest and deepest prediction market of all? In "Transformative AI, Existential Risk, and Real Interest Rates," economists argue that truly transformative AI -- something on the scale of the Industrial Revolution¹ -- should show up in long-term real interest rates. Whether AI proves incredibly beneficial or extraordinarily dangerous, those rates would rise. If markets expect AI to soon create vast, unprecedented abundance, future consumption becomes less valuable. If markets fear existential risks, future consumption holds little worth. Either scenario requires higher rates to incentivize lending. By that economic logic, I think, imminent, world-altering AI _should already be visible_ in bond markets if 2026 is shaping up as an AGI inflection point. Yet today's data suggest otherwise. Real yields are well above their ultra-low 2010s levels, but they hardly suggest markets are today pricing a sudden, trend-shattering economic rupture tomorrow. My back-of-the-envelope inference is reinforced by recent empirical work. A study by Isaiah Andrews and Maryam Farboodi examining bond-market reactions to major AI model releases finds that yields tend t _o drift down rather than rise_ following those milestones. Such movements are difficult to reconcile with the sort of large upward revisions to growth expectations that would come with imminent transformative AI. In short, markets clearly believe AI matters, just look a stocks such as Alphabet, Nvidia, and Tesla, as well as the valuations of private players such as Anthropic and OpenAI. But bond prices suggest they are treating it as driving an important, yet uncertain investment cycle -- not as the near-term prologue to human-level AI, superintelligence, or Singularity. We're not in the final countdown to Takeoff, in other words. Keep in mind: Everything you just read is utterly consistent with generative AI currently being a) an important technology with huge economic potential, both as it is and at the current pace of improvement, and b) the foundation for more explosive advances over the coming decade.² After all, any version of AI worthy of being called AGI or superintelligence during the next 7-10 years -- verification by all those betting and prediction market contracts would provide helpful confirmation for me -- would be a historic technological and socioeconomic event. Let the steeling, buckling, and preparing begin, if you wish. For me, let's just see if an AI agent can successfully do all my Christmas shopping next year. Get 50% off for 1 year 1 Maybe thinking in terms of the Industrial Revolution is too cautious. Here's the relatively cautious Google DeepMind boss Demis Hassabis on the future impact of our AI future: "It'll be 10 times bigger than the Industrial Revolution - and maybe 10 times faster." 2 Obviously definitions matter. Some claim AGI has basically arrived, pointing to systems like Claude Opus 4.5 that can now perform much of a programmer's job directly on a computer. * * * **On sale everywhere** _**The Conservative Futurist: How To Create the Sci-Fi World We Were Promised**_ | | ---|---|--- #### Invite your friends and earn rewards If you enjoy Faster, Please!, share it with your friends and earn rewards when they subscribe. Invite Friends --- | | | Like --- | | Comment --- | | Restack --- (C) 2025 James Pethokoukis 548 Market Street PMB 72296, San Francisco, CA 94104 Unsubscribe