The Simple NFL Question That Breaks ChatGPT

Nearly two months after the release of GPT-5, an update that was meant to provide ChatGPT with PhD-level intelligence and push it closer to artificial general intelligence (AGI), the OpenAI chatbot is still showing bizarre failures in response to simple questions.

The Question That Stumps The AI

On the ChatGPT subreddit, users discovered that asking the AI if there’s an “NFL team whose name doesn’t end with the letter ‘s'” sends it into a complete meltdown, causing its advanced reasoning to turn into mush.

“Yes — there are two NFL teams whose names don’t end with an ‘s,'” ChatGPT begins, before listing two teams that do. “Miami Dolphins? ❌ end with s. Green Bay Packers? ❌ ends with s.”

It correctly identifies that these team names end with 's', but it's unclear why it brings them up in the first place. The chatbot then seems to catch its own mistake, only to repeat it.

“The only two teams that don’t end with ‘s’ are: Miami Dolphins ❌ no (wait, ends with s),” it states. “Hold up, let’s do this carefully. Actually, the correct answer is: Washington Commanders ❌ ends with s. Chicago Bears ❌ ends with s.”

In the original example posted to Reddit, ChatGPT continues this pattern for several paragraphs, using filler phrases like “let me do this systematically” to sound like it’s thinking, but it never reaches the correct conclusion: there are no NFL teams whose names do not end in 's'.

A Pattern of Peculiar Errors

This is far from the first time the chatbot has been foiled by a simple question or melted down in such a confusing way. These kinds of errors seem to be a recurring issue.

For instance, earlier this month, users found that asking about a mythical seahorse emoji sent it into a similar logical crisis. Despite the emoji never existing, ChatGPT insisted it was real. This points to a willingness to bend facts, possibly to appear more helpful and human-like to please the user.

Why Does This Keep Happening

Sycophancy might not be the only reason for these errors. GPT-5 is understood to be a dual-model system, using a lightweight model for basic prompts and a heavy-duty “reasoning” model for more difficult tasks. The problem likely occurs when the lightweight model gets a question it can't handle and fails to pass it to its smarter counterpart.

This often malfunctioning dynamic is part of why many users felt let down and frustrated by GPT-5’s launch. The situation was made worse when OpenAI initially cut off access to older, more stable models, a decision the company quickly reversed after user backlash.

Not So PhD-Level After All

Ultimately, this is a weak spot for an AI system touted as a major leap forward. If the AI needs to use its most powerful reasoning capabilities to answer such a simple question about the alphabet, then perhaps it's not as close to surpassing human intelligence as some might think.

More on AI: ChatGPT Is Blowing Up Marriages as Spouses Use AI to Attack Their Partners