ChatGPT, launched in November 2022, now has 100 million weekly energetic customers.
OpenAI has acknowledged that GPT-4 LLM, which powers the ChatGPT Plus Chatbot, has obtained “lazier.” Here’s what you must know.
OpenAI was within the information for the higher a part of November when CEO Sam Altman was fired after which employed again. And whereas that will have been a short lived dangerous look, OpenAI isn’t making an attempt to lag behind in comparison with rivals, particularly with Google bringing the massive weapons with the Gemini 1.0 AI mannequin household. This is clear in the truth that the corporate has acknowledged that GPT-4 LLM, which powers the ChatGPT Plus Chatbot, has obtained “lazier.”
“We’ve heard all your feedback about GPT-4 getting lazier! We haven’t updated the model since Nov 11th, and this certainly isn’t intentional,” OpenAI posted on X (previously referred to as Twitter).
It additional added, “Model behavior can be unpredictable, and we’re looking into fixing it.”
For these uninitiated, currently, customers utilizing GPT-4 have recommended that OpenAI has degraded the expertise, and that the mannequin is nearly “fearful” of giving solutions. While some others mentioned that they have to be extra particular with their prompts, which was not wanted earlier.
In its protection, OpenAI additional notes that the mannequin has not “somehow changed itself” since November 12 (final up to date). But, the issue exists in the truth that “differences in model behavior can be subtle,” and that solely sure prompts may be “degraded.” Despite these modifications, OpenAI notes that it could take some time for customers and staff to note these patterns.
OpenAI has not promised a set deadline to launch the brand new, up to date mannequin, however it would definitely be watching rivals like Google with the discharge of the Gemini 1.0 household.
