Import AI 458: Reckoning with the future; and a singularity story
Import AI · Jack Clark · 2026-05-26
Anthropic co-founder Jack Clark, in a 2026 Oxford lecture, argues that AI progress is advancing faster than society acknowledges, predicts recursive self-improvement could begin within two years, and describes how Claude has already displaced most human coding work inside Anthropic while humans shift to verification and oversight roles.
Appears in
Extraction
Topics: ai-progressrecursive-self-improvementsingularityorganizational-changeai-forecasting
Claims
- AI systems are on a trajectory to surpass collective human intelligence, not merely individual humans, within the foreseeable future.
- Recursive self-improvement — an AI system capable of designing its own successor — could happen within two years.
- The majority of code at Anthropic is now written by Claude, with humans shifting to a verification layer atop a virtual AI workforce.
- In an internal experiment, a single human effectively managed nine synthetic research agents to conduct real research.
- A globally coordinated slowdown of AI development would likely be beneficial, but is practically impossible given commercial and geopolitical competition.
- Autonomous companies generating tens of millions of dollars in revenue could exist by November 2026.
Key quotes
One can concoct plausible scenarios by which AI could kill every single person on the planet. To think building this technology is without risk would be an act of hubris or insanity.
I believe this could happen within the next two years, and possibly sooner.
The most amazing part of this is that, to torture the analogy, the lens for the telescope I use here comes from me — specifically, from a hobby I've had for the last ten years.
The rapid advance in AI technology presents all of us with a choice: explore the future, or retreat from the present.