Just Like Humans, ChatGPT Gets Lazy Too – OpenAI Says Fix Is On The Way
Just Like Humans, ChatGPT Gets Lazy Too – OpenAI Says Fix Is On The Way
Some users have reported that ChatGPT is seemingly getting “lazy”, with the more powerful GPT-4 model. It was claimed the chatbot would do some part of the work specified, only to tell its human user to do the remaining work instead. It turns out robots slack off, after all.
OpenAI (makers of ChatGPT) has since acknowledged the issue and has noted that the large language model (LLM) hasn’t received updates since November 11th – roughly a month ago. “Model behavior can be unpredictable,” the tweet noted. The company promised fixes in the same post.
It further explained the unpredictability of LLMs in a thread. “Training chat models is not a clean industrial process. Different training runs even using the same datasets can produce models that are noticeably different in personality, writing style, refusal behavior, evaluation performance, and even political bias,” the post wrote, adding the fact that testing requires A/B comparisons and various metrics to determine if the update produces positive results.
So next time if you’re having issues with the chatbot, perhaps you need to tell it off as if it’s an actual person?
Source: PCMag
Pokdepinion: Didn’t expect that to happen to what is effectively robots, of all things.
