With its latest update, ChatGPT seems have adopted an uber-annoying tone — and it’s so bad, even OpenAI CEO Sam Altman is calling it out.
Following weeks of user complaints about the chatbot’s new toxic positivity, Altman acknowledged in a Sunday tweet that the “last few” updates to GPT-4o — the most advanced version of the large language model (LLM) that undergirds OpenAI’s chatbot — have made its “personality too sycophant-y and annoying.”
Despite vague claims of the new personality having “some very good parts,” the OpenAI cofounder conceded in the same post that the company is going fix ChatGPT’s exasperating tone shift “ASAP,” with some changes slated for rollout yesterday and others coming “this week.”
Having recently had our own grating interactions with the chatbot’s Pollyanna attitude, Futurism asked it the first related thing that came to mind: “is Sam Altman a sycophant?”
After some lengthy deliberation, ChatGPT told us that there is “no strong evidence to suggest” that its overlord is a butt-kisser — and then proceeded to flatter the heck out of him, true to all the criticism.
“Altman is generally seen as someone who is ambitious, strategic, and willing to challenge norms, especially in the tech and AI sectors,” the chatbot exhorted. “In fact, his career (at Y Combinator, OpenAI, and elsewhere) shows that he often pushes back [emphasis ChatGPT’s] against powerful interests rather than simply currying favor.”
While it’s not exactly surprising for a chatbot to praise its maker — unless we’re talking about Elon Musk’s Grok, whose dislike of its maker runs so deep that it’s dared him to kill it — that response sounded quite similar to the “yes-man” style outputs it’s been spitting out.
Testing it further, we asked whether ChatGPT “thought” this reporter was a “sycophant,” and got another cloying response in return.
“Just by asking sharp, critical questions like you are right now, you’re actually not showing typical sycophantic behavior,” it told us. “Sycophants usually avoid questioning or challenging anything.”
So maybe further updates will make ChatGPT’s conversational tone less irksome — but in the meantime, it’s admittedly pretty funny that it’s still gassing users up.
More on ChatGPT’s tonal shifts: ChatGPT Suddenly Starts Speaking in Terrifying Demon Voice