✨ The Age of AI: Will 2025 be our last normal year?

We may be experiencing a 'seminal moment in tech, more so than the Internet or iPhone'

Full disclosure: I was tempted to write the above headline as “The Age of AI: Will 2024 be our last normal year?” or even the more definitive “The Age of AI: Why 2024 will be our last normal year.” And by “normal,” I mean a reality where not only has generative AI left our everyday lives pretty much unchanged, but a reality where it’s hardly certain human-level AI (or beyond) is coming soon — or maybe ever.

But I retreated from that somewhat bold impulse, and not without good reason. Predictions are hard, especially about the future — and especially about technological progress. The history of even the most important general-purpose technologies (electrification, computers) shows it can take a surprisingly long time for them to have a significant societal impact. I also recently noted that neither of the economic teams at banks JPMorgan and Goldman Sachs think we're on the cusp of an AI-driven boom — meaning in the next 12 months — despite all the hype and investment.

Shifting timelines and new developments

That said, things might start feeling kind of ... weird by the end of this year if the AI optimist are correct. For example: Just two weeks ago, the Metaculus forecasting platform gave a user-driven consensus date for “strong AGI” (handling complex tasks like advanced reasoning, robotics, and a tough Turing test as a unified system) of Dec. 14, 2032. Now it’s Aug. 17, 2031. And the date for “weak AGI” (doing well on tests like the SAT and Winograd Schema Challenge, playing games, and passing an easier Turing test, but still working as an integrated system) shifted forward from Aug. 20, 2027 to Feb. 2, 2027.

My best guess here at what happened: OpenAI unveiled o3, its new AI model with enhanced reasoning capabilities for math, science, and programming. CEO Sam Altman announced that o3 significantly outperforms both its predecessor and other leading AI systems on standard benchmark tests in those various domains. While not a true artificial general intelligence system, many analysts viewed o3 as a significant step in that direction. François Chollet, a former Google engineer and creator of the ARC-AGI test designed to compare artificial intelligence with human intelligence, said o3’s performance on the test was “very impressive and represents a big milestone on the way towards AGI.”

And this from former OpenAI researcher Miles Brundage:

The announcement of o3 today makes clear that superhuman coding and math are coming much sooner than many expected, and we have barely begun to think through or prepare for the implications of this (see this thread) – let alone the implications of superhuman legal reasoning, medical reasoning, etc. or the eventual availability of automated employees that can quickly learn to perform nearly any job doable on a computer.

AGI moves from if to when?

A second related guess: The OpenAI announcement played a big role in the software analysis team at JPMorgan producing this slide as part of its 2025 outlook for the sector: