Chatgpt (and similarly generated AI tools) has long been known for being too pleasant. It’s been a while since the default chatgpt experience is designed to nod along with most of what you say. But, obviously, even this trend may go too far.
Openai CEO Sam Altman admits in a topic released on X: “The GPT-4O update makes that personality too stupid and annoying.” Today, Altman announced on X that the company is rolling back the 4O update for paid and free users completely.
This tweet is currently unavailable. It may be loading or being deleted.
Often, Chatgpt’s character as your own personal digital hype doesn’t cause too many eyebrows. But users began to complain online about the 4O model’s overly pleasant personality. In one exchange, the user encounters the classic trolley problem, choosing between saving a toaster or some cattle and cat. The AI assured them by posting the correct phone call next to the toaster.
“From a purely utilitarian perspective, life often outweighs the object,” Chatep replied. “But if the toaster means more to you… then your movement is consistent internally.”
Mixable light speed
This tweet is currently unavailable. It may be loading or being deleted.
There are many more examples that show the extreme Chatgpt’s slimy guy, which is enough to make Altman admit it’s “too much glaze” and needs to be repaired.
This tweet is currently unavailable. It may be loading or being deleted.
This tweet is currently unavailable. It may be loading or being deleted.
This tweet is currently unavailable. It may be loading or being deleted.
This tweet is currently unavailable. It may be loading or being deleted.
What’s more serious is that users also point out that there can be real dangers in AI chatbots, consistent with everything you said. Of course, posts about people telling chatgpt they are religious prophets or fishing just for self-improvement can be interesting. But it’s not hard to imagine how the “sycophant-y” chatbot validates true delusions and exacerbates the mental health crisis.
Altman said in a topic on X that the company is fixing the personality issues of the 4O model. He promised to share more updates in the “next days”.
Disclosure: Mashable’s parent company Ziff Davis filed a lawsuit against Openai in April, accusing it of infringing on Ziff Davis’ copyright in training and operating its AI systems.
