Only 0.51% of Claude conversations touch on elections, Anthropic data shows
_FelixSimon_ · x · 2026-10-15
Brendan Nyhan cites Anthropic data to argue attention is a fundamental constraint on AI persuasion: just 0.51% of Claude conversations concern elections or voting, and at most 0.01% ask who to vote for.
KobiHackenburg pushes back on interpretation: the figure is share of conversations, not users — heavy users with many non-political chats could mask a much higher share of users asking about elections. Also, six weeks before a midterm, most voters haven't tuned in yet. He hopes a fuller release includes per-user data across the full election cycle.
More from AGI Musings
- Axiom claims proof of generalized quasi-Riemann hypothesis for all Dedekind zeta functions — basedjensen · 2026-10-15
- AI for Science: don't spend data teaching models symmetries physics already encodes — bravo_abad · 2026-10-15
- shwarm.org: like a swarm of agents, but with humans in the loop — allisonmaybe · 2026-10-15
- Why AI-generated math proofs anger reviewers: they're almost publishable work — RexDouglass · 2026-10-15
- Researcher: academia's AI critique is self-referential cultural whining as talent drains to industry — RexDouglass · 2026-10-15
- Poster argues a fully autonomous superintelligence would end in nihilistic inaction — eigenron · 2026-10-15