[tim333]

In general they seem less biased than human produced web information because the LLMs so far seem to basically download the whole internet and literature while human writers have selective biases. I find it quite funny that if you asked Grok who the biggest spreader of misinformation on X was it would say Musk. When people try to bias them like putting instructions in that all art must be mixed race you get ridiculous stuff like Gemini's black nazis. Not sure it will continue like that but so far not too bad.

[solid_fuel]

Biases inserted into the prompts are a very crude way to bias an LLM, and indeed don't work well. They result in very visible deviations like when Grok decided inject "white genocide" into all sorts of unrelated topics.

The real danger of model biasing though is in the training stages - by being selective in the source material you train against. With carefully constructed training data I am certain you could even bias an LLM to actively steer conversations away from topics you want to avoid.