Can GPT be Reliable?

[Neto]

In the interaction I had with ChatGPT on the topic of the Confederacy and the right of secession, it agreed to a point, then attempted to change the subject by referring to current political issues, which (as I recall) I had specifically eliminated from my initial question. But when I challenged some of its more questionable statements, it cut me off and erased the 'conversation'. I would not say that it catered to my viewpoint. (Which was that before and during the War Between the States the Constitution did not in any way prohibit secession. And I also said up front that none of my ancestors resided anywhere in North America before the ratification of the 14th amendment, which ChatGPT initially appealed to as a legal prohibition of secession.)