Priming a question like "I think X—do you agree?" doesn't refine thinking, it coercively steers respondents towards replying "yes", which just creates confirmation bias and discourages dissent. Your result may be faster consensus, but it definitely won't be better judgment. Instead you'll get exactly the kind of low-quality groupthink you're claiming to fight.
Agreed. Asking "Do you agree?" or "Is this right?" will more often than not result in AI simply agree with you rather than provide critical feedback. Instead, I prompt in a way that essentially gets AI to play devil's advocate by asking questions like "What am I missing?", "How can I strengthen this point?", or "From your perspective as a [type of role], what else should I consider?”, which generates different viewpoints that may challenge my thinking, and then I can use my judgment to evaluate how valid they are and decide which insights to incorporate.
I agree with this! To be clear, I am not suggesting to ask the AI if it agrees with you, I am suggesting you ask your manager or others in the organization this question. key is to go in trying to falsify your assumptions, not trying to convince the other party (going in with the goal of “gettting to yes” will not help you improve your judgement, totally agree there)
Appreciate this post, and looking forward to your course. Have you read the HBR Guide to Gen AI for Managers yet? Thoughts on how approaches and frameworks from the book complement your course?
Does ChatGPT et al disagree ot answer that you aren't right at some point? I get your message and it's very powerful to force oneself to think a point of view before asking, but I doubt you'll receive a negative answer (as it sadly happens with many humans). I'll try your approach to find out
Yes, said this above, but I do not recommend taking this approach with the AI. I recommend taking it with your manager or others who have more context than you and also have strong judgment themselves.
Priming a question like "I think X—do you agree?" doesn't refine thinking, it coercively steers respondents towards replying "yes", which just creates confirmation bias and discourages dissent. Your result may be faster consensus, but it definitely won't be better judgment. Instead you'll get exactly the kind of low-quality groupthink you're claiming to fight.
I started replacing
“I think the answer is _____. Do you agree?”
with
“A colleague told me the answer is _____. Do you agree?”
This replaces the AI's bias to agreeing with you with a more factual answer.
Agreed. Asking "Do you agree?" or "Is this right?" will more often than not result in AI simply agree with you rather than provide critical feedback. Instead, I prompt in a way that essentially gets AI to play devil's advocate by asking questions like "What am I missing?", "How can I strengthen this point?", or "From your perspective as a [type of role], what else should I consider?”, which generates different viewpoints that may challenge my thinking, and then I can use my judgment to evaluate how valid they are and decide which insights to incorporate.
I agree with this! To be clear, I am not suggesting to ask the AI if it agrees with you, I am suggesting you ask your manager or others in the organization this question. key is to go in trying to falsify your assumptions, not trying to convince the other party (going in with the goal of “gettting to yes” will not help you improve your judgement, totally agree there)
Appreciate this post, and looking forward to your course. Have you read the HBR Guide to Gen AI for Managers yet? Thoughts on how approaches and frameworks from the book complement your course?
Does ChatGPT et al disagree ot answer that you aren't right at some point? I get your message and it's very powerful to force oneself to think a point of view before asking, but I doubt you'll receive a negative answer (as it sadly happens with many humans). I'll try your approach to find out
Yes, said this above, but I do not recommend taking this approach with the AI. I recommend taking it with your manager or others who have more context than you and also have strong judgment themselves.