OpenAI responded to the complaints that ChatGPT’s GPT-4 is getting “lazier.” The AI firm responded on X saying, “we have heard all of your suggestions about GPT4 getting lazier! we have not up to date the mannequin since Nov eleventh, and this definitely is not intentional. Mannequin conduct may be unpredictable, and we’re wanting into fixing it.”
In brief, OpenAI is saying they haven’t made any current modifications to ChatGPT or GPT-4 that will make it lazier. The truth is, no modifications have been made since November eleventh. However they defined that thee mannequin itself “may be unpredictable” and that’s what they’re wanting into repair.
OpenAI added, “to be clear, the thought just isn’t that the mannequin has someway modified itself since Nov eleventh. it is simply that variations in mannequin conduct may be delicate — solely a subset of prompts could also be degraded, and it could take a very long time for purchasers and workers to note and repair these patterns.”
Listed below are these posts:
we have heard all of your suggestions about GPT4 getting lazier! we have not up to date the mannequin since Nov eleventh, and this definitely is not intentional. mannequin conduct may be unpredictable, and we’re wanting into fixing it 🫡
— ChatGPT (@ChatGPTapp) December 8, 2023
to be clear, the thought just isn’t that the mannequin has someway modified itself since Nov eleventh. it is simply that variations in mannequin conduct may be delicate — solely a subset of prompts could also be degraded, and it could take a very long time for purchasers and workers to note and repair these patterns
— ChatGPT (@ChatGPTapp) December 8, 2023
In the meantime, additionally they posted this over the weekend:
when releasing a brand new mannequin we do thorough testing each on offline analysis metrics and on-line A/B assessments. after receiving all these outcomes, we attempt to make a knowledge pushed choice on whether or not the brand new mannequin is an enchancment over the earlier one for actual customers
— ChatGPT (@ChatGPTapp) December 9, 2023
we’re all the time striving to make our fashions extra succesful and helpful for everyone throughout thousands and thousands of use instances. so please hold the suggestions coming! it helps us keep on high of this dynamic analysis downside 🙏
— ChatGPT (@ChatGPTapp) December 9, 2023
Discussion board dialogue at X.