Political Compass of LLMs: Everyone Lands Lib-Left

An unslop.run writeup put the major LLMs through the Political Compass questionnaire and found they land, near-uniformly, in the libertarian-left quadrant. Grok is the one model that wanders — and only about half the time.

#The fine print

The Political Compass test is a self-selected 62-question instrument with no validation as a measure of model behavior, and forced-choice agreement questions are exactly the format RLHF'd assistants handle worst. Treat the result as a description of stated preferences on a specific quiz, not of how a model behaves in deployment. The HN thread (41 points, 76 comments) spent most of its energy on precisely that objection.

Sources: unslop.run, HN discussion