Sources
Simon Willison: LLMs Build Accurate Psychological Profiles from 1,000 Public HN Comments
Simon Willison demonstrates that a simple prompt — 'Profile this user' + their last 1,000 Hacker News comments (trivially fetched via the public HN Algolia API) — produces detailed, accurate psychological and professional profiles. The exercise reveals that any pseudonymous online presence with sufficient comment history is effectively deanonymizable at zero cost. Willison calls it 'mildly dystopian' and frames it as an important capability to understand as LLMs become more integrated with public data pipelines.
↳ Follow the thread