To be fair, Americans have previously asked ouija boards, psychics, and the Magic 8-Ball the same question in the past… at least they’re not asking Grok.
Yet.
Yet.
To be objective, you should test across ChatGPT, Claude, Gemini, and Grok, to see how fair they each are.
What does that mean?
More importantly, that report is more than a year ago. I would be surprised if the chatbot still calls itself that.
This seems like an irrelevant distraction.
It doesn’t call itself anything, it doesn’t have any conception of identity, nor fur that matter the semantic meaning of either part of the portmanteau it found in its training corpus… what it does have is a bias towards the bias of its trainers, in this case paid for by someone who would find it really (if pathetically) funny if their personal LLM was automatically fascist.
It’s also got a marked bias towards generative CSAM, according to investigations and filings in multiple jurisdictions… which also implies it’s best for voters not to ask it questions.
You don't know that.
Give a concrete example of a candidate where between Grok and one of the other models the lean is tilting the scales.