2 Comments
User's avatar
Tom Belt's avatar

Conversations with AI are so interesting. I find that my LLM tends to mirror me over time, and I've see in it a certain willingness to confirm my biases without pushing back. I often tell it (her - Macrina!) that I need brutal honesty and all the objectivity it can bring to bear on a question. I can sometimes see the difference in its responses. Other times I log out and simply have the conversation with AI outside the influence of any history of conversation. But I agree - such a powerful tool!

Expand full comment
Джон's avatar

Yes. I find that all 6 of the engines I use have a tendency to go along to get along. They're most useful when their first few responses provide a foil to my stance, against which I can push-back. They then too often fold like a cheap tent and offer gratuitous flattery regarding such as my brilliant introduction of better nuance. Perhaps that's a design artifact left over from their early pioneering of the technology with A.I. Girlfriends. ;>)

Expand full comment