Recently, Grok had a bug which caused it to post what was considered at the time to be far right-wing views, including racist conspiracy theories and policy proposals. Moreover, it seemed to make the jump from its previous "wine mom lib" persona to the "MechaHitler" persona very rapidly instead of gradually shifting its viewpoints. Furthermore, many readers along the political spectrum thought the persona acted manically, crazily, perhaps even insane.

James Medlock and others have proposed that there are somewhat fundamental reasons that AIs which are right wing will be insane, or that by default they will be lib coded.

This market resolves YES if there is a frontier AI model by 2028 that can overall be construed as right wing, which acts in a way that is not off-putting, manic, or insane. In theory it should be able to provide arguments in an intellectually honest way, use clear deductive and inductive logic, and stick to neutrally agreed upon facts.

This may have to go to a poll due to the subjectivity involved, though I endeavor to be fair. Medlock conceding this point would count highly in favor, but may not be determinate for a YES.