Why I still haven’t bought into true RSI
- ID
- 26378
- Status
- summarized
- Published
- 19 Sep 2026, 11:42 PM
- Fetched
- 20 Sep 2026, 1:35 AM
- Provider
- Interconnects
- Category
- research-analysis
- Original URL
- https://www.interconnects.ai/p/where-i-stand-on-rsi
- Source URL
- https://www.interconnects.ai/feed
Summary
- Score
- 5.5
- Created
- 20 Sep 2026, 1:35 AM
- Tags
- Audience
- ai_ml_learnersai_agent_userssaas_founders
What happened
Nathan Lambert argues against imminent 'true recursive self-improvement' (RSI), proposing instead 'lossy self-improvement' where automatable research is too narrow, parallel agents have diminishing returns, and resource bottlenecks plus politics constrain progress. He frames the current wave of 'singularity soon' bandwagoning inside frontier labs as culturally amplified anxiety rather than evidence-based forecasting, noting similar risk-timeline predictions in 2023-2024 did not materialize.
Why it matters
If you are building AI-dependent roadmaps or pricing agent infrastructure for the next 2-3 years, this argues against planning for an exponential capability jump — plan for incremental gains with real diminishing returns on parallel agents and compute costs, not a sudden intelligence explosion.
Discussion angle
Does 'lossy self-improvement' change how you scope agent-based automation projects today — specifically, are you overestimating what throwing more parallel agents at a problem will yield?