Geoffrey Irving Backs Sloppy Empirics Plus Principles for AI Safety

Geoffrey Irving replied that a mix of sloppy empirics and a few principles stands a better chance than empirics alone. He stated there remains some shot even if progress moves fast, because rigorous research could still produce key changes. Irving holds the role of Chief Scientist at the UK AI Security Institute. His earlier positions include work at DeepMind, OpenAI, and Google Brain centered on alignment. The comment addresses research strategies without claiming any specific outcome or broader consensus.