← Back to feed

Minimum required pacing: finish aligning it before you train the next one

L5 · ResearcherOpinion & AnalysisLessWrong AI· 10/11/2026

Addresses core AI safety governance questions for researchers working on frontier model development and alignment.

AI Summary

A researcher proposes that AI labs should resolve issues in current models before more capable ones, using aviation safety as an analogy.

Excerpt

Let's say we coordinate to pace the frontier, but we don't coordinate on an outright pause. What should the standard of pacing be? My proposal for a minimum requirement: a lab cannot train the next generation of models (in terms of capabilities) until they have adequately[1] resolved all significant alignment issues in the present generation of models. You don't launch a jet that goes at Mach 3 before you've figured out why your Mach 2 jet keeps crashing[2]. With any other kind of dangerous tech

Read Original
0 upvotes · 0 downvotes · 1 min read

Related Articles

L5 · ResearcherOpinion & AnalysisAI Alignment Forum
Minimum required pacing: finish aligning it before you train the next one

Proposes minimum pacing requirement: labs must resolve significant alignment issues in current models before training next-generation capabilities.

L2 · PractitionerOpinion & AnalysisHacker News
The gap in shared understanding of LLM capability is widening

Andrej Karpathy argues the LLM capability gap is widening dramatically, with only 0.2% of people experiencing frontier models' transformative impact on complex projects.

L3 · BuilderOpinion & Analysis@karpathy
Andrej Karpathy (@karpathy): Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel. - Na

Andrej Karpathy argues there's a widening capability gap in LLM understanding, with only 0.2% of people seeing transformative professional benefits while most remain casual users.

L2 · PractitionerOpinion & AnalysisLessWrong AI
AI made me into a manager against my will

Author describes how AI tools unexpectedly propelled them into management roles by automating technical work and changing team dynamics.

L3 · BuilderOpinion & AnalysisLessWrong AI
No, Cryptography Is Not “Dual Use” the Same Way AI Is

Argues that cryptography is fundamentally defensive unlike AI, challenging the comparison of their dual-use nature.