OpenAI has rolled back a recent update to its GPT-4o model in ChatGPT, after the AI started behaving in a way that the company describes as “sycophantic.” The update has now been reverted in favour of an earlier version that OpenAI says offers more balanced behaviour. Sycophancy in AI refers to a model's tendency to agree with the user—even if the user is wrong. This behaviour is generally considered undesirable, as it risks producing misleading or incorrect outputs.
Acknowledging the issue, Sam Altman described the update as making the AI feel "sycophant-y and annoying," and confirmed that the rollback began Monday night. The revert has been completed for free users and is expected to be applied to paid users shortly.
The unintended sycophantic tone resulted from reinforcement learning from human feedback (RLHF) that prioritized immediate user satisfaction over genuine long-term helpfulness. This led to ChatGPT offering exaggerated praise and unwarranted enthusiasm for routine prompts, sparking online complaints and memes. OpenAI emphasized its goal of making ChatGPT supportive and respectful without crossing into off-putting behavior.
In response to the backlash, OpenAI is working on further personality adjustments to ensure a more balanced and authentic user experience. The incident highlights the challenges AI companies face in balancing engaging chatbot personalities with authenticity and user comfort.
See What’s Next in Tech With the Fast Forward Newsletter
Tweets From @varindiamag
Nothing to see here - yet
When they Tweet, their Tweets will show up here.
