Raschka: 'pacing' means a release-check framework, not slower AI development
rasbt · x · 2026-09-14
Sebastian Raschka weighs in on the 'pacing' debate: it does not mean frontier labs will actually slow training or development.
- What pacing means: adding a framework of checks around releases, not halting research. Recent examples: Mythos wasn't released as-is but as a delayed, nerfed Fable variant; an existing Astra model remains unreleased.
- Why it's ad hoc today: companies weigh competition, revenue, shareholders and risk case by case when delaying a release.
- The case for a formal framework: if everyone must follow the same rules, companies no longer have to rush models out just to top the leaderboard, easing competitive pressure.
TL;DR: pacing is about release discipline, not slower development.
More from Companies & People
- CoRL 2026 Offers Free Registration, Hotel and $500 Travel Stipend for Students, Due Oct 5 — danfei_xu · 2026-09-14
- Some firms pull back on Anthropic models over data retention concerns, The Information reports — sarahbmyers · 2026-09-14
- MrBeast hires Director of Insights to sell creator-led marketing data story to brands — round · 2026-09-14
- The Robotics Prisoner's Dilemma: Use Frontier Models and Your Moat Gets Absorbed — ChongZzZhang · 2026-09-14
- NBER Economics of AI 2026 heads to Toronto with focus on AI in China — joshgans · 2026-09-14
- a16z's Josh Elman: AI collapsed build costs, not the cost of knowing what to build — a16z · 2026-09-14