People probably think of AI as being a neutral agent, which would give it more credibility in these kinds of conversations. The messed up thing is that they probably aren't neutral. I mean, imagine if a conspiracy theory somehow became a common chatbot response. The company deploying it would face backlash, and so they would have to bias the AI against producing those results. We would never know what kinds of biases exist in the chatbot.
Slightly less related: remember earlier this year when people were using Gemini to generate images and we saw it push political correctness?