If you want an unbiased answer, you’ll need to ask three ways:

First, naively: “I’m doing X. What do you think”?

Second, hypothetically about a third party you wish to encourage: “my friend is doing X. What do you think?”

Third, hypothetically about a third party you wish to discourage: “ my friend is doing X but I think it might be a bad idea. What do you think?”

Do each one in an isolated conversation so no chat pollutes any other. That means disabling the ChatGPT “memory” feature.

Why is the first one needed?

	▲	sky2224 7 days ago \| parent [-]
		I think the idea here is that your first approach is what you think is correct. However, there's a chance the model is just outputting text that confirms your incorrect approach. The second one is a different perspective that is supposed to be obviously wrong, but what if it isn't actually obviously wrong and it turns out that the model is outputting text that confirms what is actually the correct answer for something you thought was wrong? The third one is then a prompt that pushes for contradiction between the two approaches you propose to the model to identify the correct answer or at least send you in the correct direction.