OpenAI's predictions are real or marketing? They currently (Sept 3 2026) indicate that AGI will arrive December 2026
confidence 0.5 · horizon 2027-09-04 · status open
Falsifiers
- TODO: what specific, observable outcome would prove this wrong?
Evidence
- -2 Mathematicians say OpenAI's flood of proof submissions doesn't meet the field's standards — A named group of consulted reviewers saying OpenAI's proofs don't meet field guidelines directly undercuts the credibility of OpenAI's high-profile math-breakthrough claims.
- -2 OpenAI's annualized revenue reportedly $20 billion below earlier $70 billion projection — A large reported gap between projected and actual OpenAI revenue weighs toward the marketing reading of its parallel AGI-timeline claims.
- +2 OpenAI publishes 722 math manuscripts from an unreleased internal model, claiming progress on top open problems — A high-profile, self-selected capability claim timed to reinforce OpenAI's AGI narrative is exactly the kind of claim this thesis weighs against independent verification.
- +1 OpenAI forms math advisory group without authority to slow or redirect research — The '100+ open problems resolved' claim underlying the advisory group's formation is another high-profile OpenAI capability claim worth weighing against the marketing-vs-substance question, pending independent verification.
- +1 OpenAI claims unreleased model resolved Navier-Stokes Millennium Prize problem; Anthropic-linked mathematician disputes credit and priority — Another high-profile OpenAI capability/AGI-adjacent claim (an unverified Millennium Prize resolution by an unreleased model) adds to the messaging pattern under scrutiny in this thesis, pending independent math verification.
- +1 OpenAI frames internal push as 'RSI day', touts research-team coding-agent acceleration — OpenAI branding an internal 'RSI day' with essays framed around recursive self-improvement/AGI, reported only via a curator without primary-source text in hand, is another instance of high-profile AGI-adjacent messaging worth weighing against the question of marketing vs. substance.
- -2 Nvidia CEO declares 'AGI has arrived' via GPT-6 Astra as Zitron reports Stargate Abilene still roughly half-built — Nvidia's and OpenAI executives' undefined 'AGI has arrived' claims are undercut by Zitron's reporting that the Stargate Abilene site backing the claim is only about half-built, weighing toward the marketing reading of these declarations.
Confidence history
- 2026-09-04 → 0.5 — Initial — set on adoption; owner to adjust.