128 · Tier 3 — High-impact political/media nodes
G. Elliott Morris
Founder and editor, Strength in Numbers
AI position
No x-risk mention
All findable AI-related writing from Morris concerns AI's use in survey research/polling (synthetic respondents, LLM-coded open-ends); no public statements found engaging with AI existential risk.
Why influential
Shapes campaign strategy and political narratives through polling interpretation, ratings, forecasting and electoral analysis. Primary policy lanes: polling; forecasting; elections; public opinion.
Who pays attention
Campaign managers, pollsters, political reporters, donors, forecasters and politically engaged data readers
Work samples
- AI Can't Replace Polling
Sep 2, 2025 · newsletter
Argues LLM-based 'AI polls' (e.g. from startup Aaru) betray the public and produce inaccurate substitutes for real survey respondents.
- Your Polls on ChatGPT (Verasight white paper / synthetic sampling report)
Sep 8, 2025 · article
Co-authored white paper testing LLM-generated 'digital twin' respondents against real polling data; found systemic errors, though within ~2pts on some questions.
- What does artificial intelligence know about polling? (episode #208)
Dec 11, 2022 · newsletter
Early piece exploring what AI/LLMs understand about polling methodology.
- Why we trust real polls over prediction markets
Aug 1, 2024 · newsletter
Compares traditional polling methodology against prediction markets as forecasting tools.
- X post announcing 'Your Polls on ChatGPT' white paper
Sep 8, 2025 · social
Announces synthetic sampling research on X, summarizing findings on LLM-based polling.
AI-world contacts
Tech-side people they have hosted, quoted, covered, or engaged — collected by the research pass. None documented yet: this is the white space.