A 2026 review of AI 2027's 2025 predictions finds that quantitative AI progress is occurring at roughly 65% of the predicted pace, with most qualitative predictions remaining on track. If progress continues at this rate, full coding automation and AI takeoff would occur from mid-2028 to mid-2030, with notable underperformance in SWEBench improvements and AI R&D automation, offset by stronger-than-expected revenue growth.
The author argues that OpenAI's charter contains a self-sacrifice clause committing to halt competition if competitors achieve AGI first, yet Sam Altman's publicly stated AGI timelines have continuously accelerated (from ~2033 in 2023 to ~2025 in recent statements) while OpenAI's GPT-5.4 trails competitors like Claude and Gemini in capability benchmarks, suggesting OpenAI should be joining forces with competitors per its own stated policy but continues racing instead.