Sebastian Raschka: 'pacing' means more checks, not slower AI development
rasbt · x · 2026-09-15
ML researcher Sebastian Raschka offers his take on the 'pacing' debate: it doesn't mean frontier labs will actually slow training, but rather add a framework of more checks. He points to recent precedent — a frontier model shipped delayed and nerfed under a different name. He adds that mild regulation creates an attractive deflection incentive: companies can say 'sorry it happened, but we followed the framework' to deflect blame and cope with anti-AI sentiment.
Related event: Raschka: Pacing Means Checks, Not Slowing Down(2 posts)→
More from AGI Musings
- Europe must bargain for frontier AI access as its models can't catch up, researchers argue — NinaDSchick · 2026-09-15
- Sociologist estimates 160 million workers power AI's hidden labor force, mostly in Global South — AntonioCasilli · 2026-09-15
- Frontier labs will pace AI progress by spending more compute on alignment and evals, says investor — GavinSBaker · 2026-09-15
- AI raised the security floor, but elite intrusions still need strong teams, researcher says — nptacek · 2026-09-15
- Accelerationists celebrate as Dario Amodei gets 'mogged' over slowdown stance — ccerrato147 · 2026-09-15
- Why Pushing AI Out of Utility Grids Is Easier Than You Think: Regulated Returns Explain It — LiberataJoystar · 2026-09-15