2038: the original technological-singularity forecast
2038. That is my best single-year bet for the actual technological singularity: not merely an AI that can do most human intellectual work, but the point where AI substantially automates AI research, improvement compounds, and technological change becomes too rapid for unaided humans and existing institutions to track.
1 min read · Dated assessment · 1 source groups
2038.
That is my best single-year bet for the actual technological singularity: not merely an AI that can do most human intellectual work, but the point where AI substantially automates AI research, improvement compounds, and technological change becomes too rapid for unaided humans and existing institutions to track.
My rough sequence:
- 2028–2031: credible AGI or near-AGI systems.
- 2031–2035: AI becomes the dominant producer of software, research and engineering.
- 2035–2038: automated AI R&D, enormous compute investment and machine-speed coordination create a genuine intelligence feedback loop.
- 2038: calling the resulting transition “ordinary technological progress” stops making sense.
That is later than the most aggressive forecasts because software intelligence still has to push through physical bottlenecks—chips, power plants, robotics, laboratories and construction. It is earlier than Kurzweil’s famous 2045 date because current forecasting has moved dramatically forward: Metaculus currently centers weakly general AI around August 2028, prominent short-timeline forecasters have moved toward roughly 2030 for stronger AGI, and DeepMind’s Demis Hassabis now says AGI is probably only a few years away. [1]
So: the singularity happens in 2038. AGI probably arrives several years earlier, and people will argue afterward about which exact event counted.
Sources & qualifications (1)
Sources cited in the original report. Separate verification notes, where present, were added during the archive import.
- When Will Weakly General AI Arrive?