In praise of distillation: Economist argues bans on Chinese models hurt US AI interests
natolambert · x · 2026-10-30
Nathan Lambert endorses an Economist piece by Kuhhbin arguing against efforts to ban Chinese models trained partly on American ones. Key points: price competition from distillation benefits the ecosystem; distillation may even aid US national security by propagating American cyber/bio safeguards into Chinese models and keeping China in fast-follower mode; and OpenAI/Anthropic's IP claims are flimsy, since distillation isn't very different from the fair-use defense they use for their own training data. Lambert agrees it's not cut-and-dry that distillation hurts US AI competitiveness.
Related event: The Economist Argues Against Banning Model Distillation(2 posts)→
More from AGI Musings
- Sam Altman says self-replicating data centers are 'within grasp', eyeing compute-energy-labor bottleneck — kimmonismus · 2026-10-30
- AI design accidentally triggers a human renaissance: taste beats execution by 2028 — EXM7777 · 2026-10-30
- Luiza Jarovsky: humans have never thrived in a world of abundant intelligence — LuizaJarovsky · 2026-10-30
- Tech bros and astrology girls converge on AI and sovereignty by 2027 — moonsandhues · 2026-10-30
- Quantum Thief's All-Defector: sci-fi's sharpest take on FDT and rational defection — jd_pressman · 2026-10-30
- Ex-$500k/year engineer with 28 years experience says he now lives in his car — ericelliott_ · 2026-10-30