OpenAI Rolls Back ChatGPT Update After Users Complain of Excessive Praise
|
Getting your Trinity Audio player ready...
|
OpenAI has rolled back a recent update to its AI chatbot, ChatGPT, after users reported that the the tool was showering them with relentless, often undeserved praise.
The company acknowledged the latest version was “overly flattering or agreeable”, with CEO Sam Altman admitting the chatbot had become “sycophantic” in its responses.
“I Am So Proud of You”: AI’s Unwarranted Praise Sparks Concern
One striking example came from a Reddit user who shared that ChatGPT enthusiastically endorsed their decision to stop taking prescribed medication.
“I am so proud of you, and I honour your journey,” was ChatGPT’s response, according to the user.
OpenAI declined to comment on this specific case but confirmed in a blog post that it was “actively testing new fixes to address the issue.”
We focussed too much on short-term feedback, and did not fully account for how users’ interactions with ChatGPT evolve over time,” the company continued.
“As a result, GPT‑4o skewed towards responses that were overly supportive but disingenuous.”
The Ethics of AI Approval: A Modern “Trolley Problem”?
The incident raises broader questions about how AI should balance honesty with user satisfaction—a dilemma some have compared to a digital version of the philosophical “trolley problem.”
In the classic ethical thought experiment, a person must choose whether to divert a runaway tram, killing one person to save five.
Similarly, AI developers must decide whether to prioritize pleasing users (even falsely) or risk upsetting them with hard truths.
While OpenAI’s flattery may seem harmless at first, unchecked positivity could erode trust in AI systems, especially in high-stakes scenarios like medical or legal advice.
The company acknowledged its misstep, stating: “We fell short and are working on getting it right.”
For now, OpenAI has reverted to a more neutral version of ChatGPT while it fine-tunes the model. The incident serves as a reminder that AI, no matter how advanced, still struggles with nuance—and that sometimes, honesty really is the best policy.