Priority AI risks
data_society's bookmarks 2026-06-08
Summary:
A happening, methinks, more than a community post (given number of authors). Fac Assc Mark Esposito writes, "I wanted to share a study I contributed to as one of 272 co-authors, just released by MIT FutureTech: Prioritizing the Risks from Artificial Intelligence.
Using the Delphi method across a genuinely international expert panel, the study asks which AI risks are most severe, which sectors and populations are most vulnerable, and where responsibility actually sits. The findings have direct relevance for governance and policy: 18 of 24 risk domains carry at least a 10% probability of catastrophic outcomes by 2030 under current trajectories, and the experts most consistently point to a structural accountability gap -- those bearing the greatest risk are not those with the greatest power or obligation to address it. Bridging that gap requires not just better technical safeguards but enforceable rules, liability frameworks, and international coordination on cross-border risks.
For a community working at the intersection of technology, law, and the public interest, I think this study offers both useful empirical grounding and a productive provocation."