Scott Alexander said, in a conditional aside, that he expects a singularity in the next few decades.

This is a hedged prediction that a 'singularity' will happen within 'the next few decades' of December 2024. Read 'few decades' as about three decades, so the horizon is the end of 2054. Count as a singularity a widely recognised technological singularity: AI that broadly exceeds human ability across most cognitive domains and drives rapid, self-reinforcing technological and economic transformation, as described by mainstream reporting and expert assessments. Narrow AI advances do not count, however impressive. TRUE: by the end of 2054, credible sources broadly agree that such a singularity or intelligence explosion has occurred. MOSTLY TRUE: it clearly occurred by the horizon, but some sources date its onset or describe its features a little differently. MIXED: by the horizon there is general or superhuman AI with large, fast-growing impact, but experts genuinely disagree about whether this amounts to a singularity. MOSTLY FALSE: transformative-AI milestones were reached only partly by 2054, or a singularity-level transition happened only after the horizon. FALSE: dated assessments around 2054 show no singularity has happened, with AI progress and economic growth not rapid or self-reinforcing in the claimed sense. The fact that Scott Alexander said he expected it is not at issue; the outcome is what gets judged.

Set before the research began.

1. E1 Experts say we are not yet in the AI singularity University of Sydney (republished from The Conversation) · 31 July 2026

   “Mathematician and science-fiction author Vernor Vinge defined it in 1993 as a point at which machine intelligence exceeds human intelligence and begins improving itself, triggering an acceleration so rapid that humans can no longer predict or control it.”

2. E2 Experts say we are not yet in the AI singularity University of Sydney (republished from The Conversation) · 31 July 2026

   “These AI models cannot change (or “learn”) while running. The model that broke into Hugging Face was identical afterwards to what it had been before.”

3. E3 Have we reached the AI singularity? The Week · 24 August 2026

   “OpenAI CEO Sam Altman believes that we have reached the singularity, an inflection point at which humankind can no longer keep up with how quickly artificial intelligence improves itself.”

4. E4 The Economics of Recursive Self-Improvement arXiv (authors affiliated with METR, Stanford, MIT, Yale and others) · 13 September 2026

   “A back-of-the-envelope calculation suggests that feedback loops are not currently strong enough to generate a self-sustaining acceleration, though they appear to be strengthening.”

The claim predicts a singularity by about the end of 2054, and today is November 2026. Evidence from 2026 shows experts saying a singularity has not yet arrived (E1, E2) and AI feedback loops not yet strong enough to drive self-sustaining acceleration (E4). Others, such as Sam Altman, say it has already been reached (E3). None of this settles whether a singularity will happen before the 2054 horizon, so it is too early to judge.