Head of Policy · OpenAI · 2020
Left to co-found Anthropic, where he leads policy work.
The seventh annual AI Index report — authored by Nestor Maslej and a fourteen-person team at Stanford's Human-Centered AI Institute, with Jack Clark serving as steering-committee co-director — provides a comprehensive, data-driven assessment of the state of artificial intelligence across research output, technical performance, responsible AI practices, the economy, education, policy, and public opinion worldwide. The report documents the accelerating pace of AI development, noting that industry has overtaken academia as the primary driver of frontier model development and that the costs of training leading systems have risen dramatically into the hundreds of millions of dollars. It tracks the growing number of AI-related regulations worldwide, the expansion of AI adoption across economic sectors, and shifts in public sentiment that show increasing awareness of both AI's potential benefits and its risks. The report also highlights significant gaps in responsible AI practices, including the lack of standardized benchmarks for evaluating safety and the limited transparency from leading developers about their training data and evaluation procedures. As a widely cited reference document used by policymakers, journalists, and researchers, the AI Index plays an important role in grounding public discourse about AI progress in empirical data rather than speculation.
An essay adapted from Clark's remarks at The Curve conference in Berkeley. Clark argues that too many people are pretending that AI cannot threaten humanity, likening today's systems to 'true creatures' revealed when the light is switched on rather than imagined shapes in a dark room. He pairs genuine optimism about AI's productive potential with a call for 'appropriate fear' of increasingly autonomous, unpredictable systems, and for honesty about both.