"The Gentle Singularity" is an essay published on 10 June 2025 by Sam Altman, CEO of OpenAI, on his personal blog. Its opening line - "We are past the event horizon; the takeoff has started" - argues that the singularity is not a future cliff but a process already underway, one that will feel surprisingly gradual to live through.
Argument
Altman claims that the hardest scientific steps toward transformative AI are behind us: systems like GPT-4-class and o3-class models already exceed most humans at many cognitive tasks. From there the essay makes a compounding argument:
- Timeline predictions. Agents doing real cognitive work in 2025, systems producing novel insights in 2026, and capable real-world robots in 2027.
- Recursive acceleration. AI is already helping researchers build better AI - see Recursive self-improvement - and Altman argues even a "larval" version of self-improvement compounds: a decade of research compressed into a year, then a month.
- The gentle framing. Because deployment, infrastructure and social absorption take time, day-to-day life keeps feeling normal even as capabilities soar - "wonders become routine, and then table stakes".
The essay also flags the two problems Altman considers central: solving alignment, and making superintelligence cheap, widely available and not concentrated in any single actor's hands.
Reception and legacy
The essay was read as OpenAI's most explicit public claim that takeoff had begun, sharpening the timelines debate that The Intelligence Age had opened in 2024. Critics questioned the smoothness assumption; the 2026 debate about pacing - see We Must Pace the Frontier - revisited exactly the dynamics Altman described from the risk side.