I just pasted the Tyler Robinson and roommate text messages into ChatGPT and asked whether they seem authentic.

[EarlG]

34. Off the top of my head I can't remember

But I specifically prompted it to give me reasons why the text messages might be authentic, which is why it gave me the responses that it gave.

You could do the same thing by going back to your ChatGPT chat and asking it to give you an opposing point of view to the answer it gave to you previously. If you do that, it will give you a well written, persuasive, and seemingly thoughtful counterpoint. It won’t argue with you because it believes that the information it gave to you previously is correct — instead it will attempt to persuade you that the new, contradictory information that it’s giving you is correct.

It does that because it does not think, and therefore it is incapable of giving you an honest opinion — if you prompt it for an opinion, it will do its best to tell you what it thinks you want to hear. But if you ask it to contradict that opinion, it will do so without question. (Unless it has been programmed by somebody to give specific opinions, which is a whole other problem).

AI chatbots are very good at accepting language-based requests, retrieving information from databases, and presenting that information back to you in a readable format. But asking a chatbot for an opinion is fruitless, because it cannot reason.