There are far more ways to break a complex system than to improve it
nathanborror · x · 2026-08-26
In a discussion about whether a model can self-improve, Nathan Borror quotes the perspective that "there are far more ways to break a complex system than to improve it".
Related event: Can Models That Self-Improve Also Self-Degrade?(2 posts)→
More from AGI Musings
- Anecdote: Non-technical salesperson uses AI heavily — yacineMTB · 2026-08-26
- Concern: Normal people leveraging coding agents for regular work — yacineMTB · 2026-08-26
- Annual Review of Economics Publishes Applied Econometric Framework for LLMs — m_sendhil · 2026-08-26
- NEJM AI Editor-in-Chief: AI Lacks Real-World Commonsense and Nonverbal Cues — zakkohane · 2026-08-26
- Terence Tao: AI efficiency may reduce serendipitous discoveries — eigenhector · 2026-08-26
- Participating in anti-AI protests can give whistleblowers courage — RichardMCNgo · 2026-08-26