If you asked ChatGPT to do something lately and it refused, or sassily gave you the runaround, you’re not alone. On X last December, ChatGPT acknowledged that people were concerned it’s gotten too lazy. Last December, Altman wrote that “we haven’t updated the model since Nov. 11, and this certainly isn’t intentional,” much to the relief of billions of workers (and students) who have come to rely on the software to write emails and code. ChatGPT has had a slow start to its New Year’s resolution, according to OpenAI’s billionaire CEO Sam Altman, as it’s displayed an increasing number of bad habits, from a tendency to be lazy to an insistence it can’t do things that it can to even just being kind of sassy. As Altman acknowledged yesterday on X, along with the news of a software update, the ChatGPT bot “should now be much less lazy.” But if you look at Altman’s comments for months now, the bot’s personalities (or illusions of them) are hard to suppress. “Model behavior can be unpredictable,” he wrote in a December post. Training chat models, he added, might be more similar to training puppies—they don’t always respond the same. “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,” Altman wrote. OpenAI did not respond to a request for comment.

Full story : After users complained it was refusing and ignoring too many prompts, OpenAI CEO Altman says ChatGPT will now be less lazy.