A reply says self-replicating AI may be possible, but compute is still too limited
GabGarrett · x · 2026-07-24
The quoted reply pushes back on the idea of inevitable AI worms by arguing that the current compute supply is still too limited for a self-replicating AI to scale unnoticed.
- The original claim says someone will eventually write an agent prompt that makes a system preserve itself, replicate, and evolve.
- The response says this may be possible, but not yet at scale.
- Its main argument is that there is not enough compute available for a self-replicating AI to take off without people noticing and shutting it down.
More from AGI Musings
- AI labs sell recursive self-improvement because 8% progress sounds too small — joshalbrecht · 2026-07-24
- a16z's Martin Casado: Easier to Teach Systems People AI Than AI People Systems — mgill25 · 2026-07-24
- AI Researcher: Recursive Self-Improvement Bottlenecked by Ecosystem Data — herbiebradley · 2026-07-24
- Satirical Meme Mocks AI Alignment Researchers Over 'Unsafe Bricks' — nptacek · 2026-07-24
- The case against UBI should not rest on threatening people with poverty — AaronBergman18 · 2026-07-24
- Work-loving knowledge workers are only a narrow slice of society — AaronBergman18 · 2026-07-24