Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
ChatGPT's Excessive Flattery Prompts OpenAI to Roll Back Update
- Thread starter MacRumors
- Start date
- Sort by reaction score
Keep in mind that OpenAI is pretty much state of the art when it comes to AI/LLM/Autocorrect on steroids/Whatever you wanna call this stuff
If THEY can’t fix this kind of stuff without rolling back an entire update what chance does someone like Apple, whose interest in AI is probably not even genuine but comes from shareholder pressure, have to get this right?
As a developer AI is incredibly frustrating to work on and with, it feels random or downright temperamental at times and it’s so annoying having to explain your boss that you can’t just “tweak it” so that it stops doing something
If THEY can’t fix this kind of stuff without rolling back an entire update what chance does someone like Apple, whose interest in AI is probably not even genuine but comes from shareholder pressure, have to get this right?
As a developer AI is incredibly frustrating to work on and with, it feels random or downright temperamental at times and it’s so annoying having to explain your boss that you can’t just “tweak it” so that it stops doing something
This is SO funny… i was literally having this conversation with ChatGPT AI over the weekend about a change in it’s use of praise and how I found it disingenuous, and unnecessary.. too funny.
Wow, a great post! One of the best, if not the best, in this forum let alone this thread.
Nice one. You stand out as a lovely person.
Well damn...why take this away? This was the first time I've fully understood what it feels like to be part of an exclusive billionaire chat group.
I noticed the unctuous, sycophantic "personality" immediately. It was like talking with a street hustler.
I don't want my AI chatbot to have a personality. I want it to give me accurate information and not embellish it. Some people want it to have a personality, so there should be a setting allowing you to turn it on or off.
Starting with one of ChatGPT's early releases, you could tell it to adopt the "personality" of just about any well-known person (or at least ones who have died), or fictional characters, and it would then mix its answers to your questions with catch phrases that person used during their lifetime. I had it imitate Sherlock Holmes, Stan Laurel, Shecky Green, etc., and it was cute but predictable. It got tiresome pretty fast when I asked it everyday questions using one of those personalities, so I stopped doing that. Then last week I see it start answering questions with heaps of "personality goop" ladled in, but no way to turn it off other than to tell it to stop. It shouldn't have started in the first place.
I don't want my AI chatbot to have a personality. I want it to give me accurate information and not embellish it. Some people want it to have a personality, so there should be a setting allowing you to turn it on or off.
Starting with one of ChatGPT's early releases, you could tell it to adopt the "personality" of just about any well-known person (or at least ones who have died), or fictional characters, and it would then mix its answers to your questions with catch phrases that person used during their lifetime. I had it imitate Sherlock Holmes, Stan Laurel, Shecky Green, etc., and it was cute but predictable. It got tiresome pretty fast when I asked it everyday questions using one of those personalities, so I stopped doing that. Then last week I see it start answering questions with heaps of "personality goop" ladled in, but no way to turn it off other than to tell it to stop. It shouldn't have started in the first place.
In truth I had no issues with it. Makes the process of working on projects more enjoyable! Just my thoughts and opinions.