Anthropic opens a can of difficult questions about AI

Anthropic is launching an initiative called "hard questions," which invites the public to submit their most thorny questions about artificial intelligence: its effects on employment, society, and families; its potential for science and medicine; or the direction that one of the most powerful technologies in history is taking.

The approach is rooted in the company's status as a Public Benefit Corporation, whose stated mission is to secure the benefits of advanced models while mitigating their risks. It extends an ongoing listening effort: a survey, the Anthropic Public Record, the first round of which polled 52,000 Americans on their hopes and fears; a survey of 81,000 Claude users across 159 countries and 70 languages; in-person focus groups; and the study of real-world Claude usage through anonymized data.

In exchange for the questions received, Anthropic says it commits to publishing and tracking the concrete actions it takes to address them, while also indicating areas where it might fall short of its goals. The initiative builds on pre-existing internal structures, including the Anthropic Institute, dedicated to the societal challenges of AI, and the Long-Term Benefit Trust, which is meant to provide unbiased oversight of the public interest mission. The public is directed to a dedicated website to view questions already asked and to submit their own.