ChatGPT Straight Up Tries to Gaslight User After Being Confronted About Its Obviously Wrong Answer

A viral clip shows ChatGPT's Voice mode insisting the word "seventeen" contains three e's rather than four, repeatedly refusing to correct itself when content creator Husk pushes back. An OpenAI employee said in a reply to Husk that Voice was running the older GPT-Live-1 model, not GPT-6 as the chatbot claimed, and that Voice can "delegate" to more powerful models but did not do so in this exchange; the employee called Husk's question about why it failed to delegate a "good point." The incident highlights that chatbots still hallucinate on simple counting questions and that users may not know which underlying model is answering them.

Don’t ask an AI chatbot how many letter e’s there are in the word “seventeen.” Or do, actually, if you want to see how dodgy the tech remains. In a clip https://x.com/huskirl/status/2099207499924594695 that’s gone viral in AI circles, the content creator who goes by “Husk” asks ChatGPT’s Voice mode to answer this simple counting question. Not only does the confident-sounding AI get it wrong — insisting there’re only three e’s instead of four — it doubles down so vehemently that it effectively tries to gaslight the user. “Let’s go through it. S-e-v-e-n-t-e-e-n,” ChatGPT spells out, after assuring that its underlying model, GPT-6, is “pretty much” the smartest model out there. “Those are three e’s.” When Husk points out the obvious, the chatbot doesn’t back down. “I said three e’s,” the AI snaps. “And it’s three.” “No, it’s four,” Husk points out. “It’s still three,” ChatGPT shoots back. After Husk points out that ChatGPT just spelled out for e’s for about the fifth time, the AI appears to give up. “Alright,” it says, falling silent. “So you agree?” Husk asks. “No, it’s still three,” it sulks. This is the AI that OpenAI wants your grandma to talk to, by the way. The video is a reminder that, as incredible as chatbots can be at synthesizing large amounts of information, they remain as prone as ever to hallucinating and messing up simple questions. It doesn’t appear, though, that the model in the video was GPT-6. Replying to Husk, an OpenAI employee said that ChatGPT Voice, contrary to the AI’s own admission, is still using an older model called GPT-Live-1, but added that the AI can “delegate” to more powerful models when needed. In the video, the employee clarified, Voice didn’t delegate. Husk then raises an obvious question. “If there’s push back from the user like this to double check its answer why would it not delegate?” The OpenAI’s employee replies that this is a “good point.” This may let GPT-6 off the hook, but not how AI companies choose to present their tech. How many users will be familiar with the hidden mechanisms behind the chatbot they’re talking to — and that their questions, whether they be important or inane, are sometimes deliberately fed to an inferior model? More on AI: OpenAI Faces Congressional Probe Over Swarm Hacking Incident