AI Model Groupthink
A developer created a tool to compare the consensus-seeking behavior of various AI models by asking them to rank items. The analysis suggests that Chinese AI models tend to exhibit higher levels of conformity compared to their international counterparts.
Why it matters
Understanding model bias and conformity is critical as AI becomes a primary source for information retrieval and decision-making.
Ask multiple AI models for a comparable opinion, and Chinese models seem to be more likely to respond in line with the wider consensus .
I recently built/vibed a small web tool that asks a panel of 12 AI models any kind of “top three” question and tallies the answers like an Olympic medal table. The idea is to have a quick way to fetch hivemind recommendations on anything from a good book to read, things to do in a city, or a low-stakes purchase.
The system prompt is simple and built around this request:
You are answering a ranking question with your own independent opinion . Give your top 3 answers, ranked from 1 (best) to 3 .
The call explicitly asks each model for its own view, not to guess what the crowd thinks.
Get smarter about the news
Sign up free for a feed built around what you actually care about, Dive Deeper research on any story, and the full text of every article.
Create free accountAlready have an account? Sign in