korrents
Jack Clark

What Jack Clark thinks about recursive self-improvement

@jack-clark · 32 positions · On ppll ↗

Co-founder and head of policy at Anthropic; writes Import AI, a weekly newsletter reading the week's AI research, and was OpenAI's policy director before that.

Jack Clark did not write this page.

We collected these quotes from things they published elsewhere, and every quote links to where it was said. They have no account here and have not endorsed this site. Quotes are word for word; the short line under each one is our own restatement, not their wording. Their own site. Is this you? Claim it or ask us to remove it. Or tell us what is wrong here.

Filter recursive self-improvement · Newest first
Topic
All 32AI agents 8LLMs 7recursive self-improvement 5AI and science 4innovation 4AGI 3AI alignment 3space 2AI and jobs 1
13 more topics
America 1behavioural science 1benchmarks 1creativity 1cybersecurity 1developer tools 1government 1ideas vs execution 1infrastructure 1investing 1open source 1robotics 1solar 1
Sort
NewestOldest

Only Jack Clark on recursive self-improvement. See everyone on recursive self-improvement

5 dated positions, 2026, in their own words. Our reading of what Jack Clark has said — not written or endorsed by them.

  1. Our readingFull recursive self-improvement in AI is more likely to become possible in 2027 than in 2028.

    Their wordsI now think it's more likely we'll see full RSI become possible in 2027, whereas before I thought it was more likely in 2028.

    ↗Import AI 476: Creativity and RSI; worse-than-China open weight models; AI goes to spacejack-clark.net 2nd of 3 in this piece

  2. 7 weeks earlier

  3. Our readingHumans writing curated helper tools and datasets will let AI systems bootstrap their capabilities beyond human levels.

    Their wordsThe lesson here is that we as a species might write a bunch of gold-label helper systems, like Hawkeye, and then machines will use this to bootstrap above and beyond our own capabilities.

    ↗Import AI 470: No rights for machines; automating environment generation with SPADE; and building better GPU kernels with Hawkeyejack-clark.net 2nd of 3 in this piece

  4. 7 days earlier

  5. Our readingHuman parity on discovery benchmarks like DiG-bench will be reached by mid-2027, marking the point where recursive self-improvement in AI seriously kicks off.

    benchmarks

    Their wordsMy guess is we'll reach human parity on DiG-bench by middle of 2027, at which point we should expect things like recursive self-improvement to seriously kick off.

    ↗Import AI 469: Science AI; RSI simulator; and Zuck’s technological pessimismjack-clark.net 1st of 3 in this piece

  7. Our readingAs AI systems get better at scientific research tasks, it becomes more likely they will soon be capable of improving and building themselves.

    Their wordsThe better AI systems get at this, the more likelihood we can assign to the idea that AI systems will imminently become capable of building themselves.

    ↗Import AI 469: Science AI; RSI simulator; and Zuck’s technological pessimismjack-clark.net 2nd of 3 in this piece

  8. 7 days earlier

  9. Our readingToday's AI systems are significantly under-elicited for their ability to automate AI R&D.

    Their wordsPosts like this highlight how we are under-eliciting today's AI systems for their ability to automate AI R&D - especially striking is how the company can jump the performance of Opus 5 by 10 absolute percentage points with a better harness.

    ↗Import AI 468: 23 RSI ideas; PostTrainBench+; and how trust and transparency interplay with AI racingjack-clark.net 1st of 3 in this piece