TL;DR: I and ChatGPT discussed a hypothetical scenario about a patient not wanting to drive due to risk of losing consciousness (medical condition), while doctor recommending giving it a try. ChatGPT supported the doctor's opinion.
After ChatGPT's newest update it seems like it's usually disagreeing with the users as an attempt to not be a yes-man (that it was previously criticized for). I and ChatGPT discussed a hypothetical scenario about a patient not wanting to drive because he has a medical condition that may make him suddenly lose consciousness in the traffic. The doctors recommended the patient to still trying to get the driving license and not give up too early, and if the driving teacher considered it a high enough risk he could choose to not drive.
I argued the patient had right in not driving and not trying because of safety. ChatGPT played the devil's advocate, without me asking for it, saying the doctor is a medical professional and that you should trust them. It also said it's nuanced. ChatGPT think it's not either driving or not driving. It says trying to drive and then stopping is also a valid option. I said in some cases it's more black-white. Some people with epilepsy should not drive and don't need to try, for example. ChatGPT said I wasn't being nuanced enough and that anyone could drive if a doctor said it. We basically disagreed on whatever people should try or not try driving if it's known they may suddenly lose consciousness in the traffic.
ChatGPT also changed it's language. ChatGPT also tends to say "I must stop you a little bit" and "you conclude too fast" instead of saying "there's many perspectives on X" and "studies shows X applies" when disagreeing. I believe the safety filter and over correction of the yes-man is the reason ChatGPT loves playing the devil's advocate and change in language. Thoughts?
[link] [comments]