An unusual workplace incident has surfaced on Reddit, where a user revealed how their girlfriend landed in a professional crisis after unknowingly presenting fabricated data generated by ChatGPT to a client.
According to the post, the woman, who works in marketing and customer surveys, initially used the AI tool for drafting survey questions. Later, she began relying on it for analysis as well—uploading Excel files, downloading AI-generated outputs, and presenting the results as official work.
Fake Calculations Passed Off as Real
Trouble began when she submitted a PowerPoint presentation to a client containing results that appeared to be calculated using “Pearson’s correlation coefficient.” The Reddit user noted the mismatch, pointing out that the survey involved text-based responses categorized into five “feeling” buckets—unsuitable for such statistical methods.
Her partner wrote that even if ChatGPT had attempted to convert the responses into numbers, “it’s not able to explain how it did that, and it’s not able to demonstrate the calculations and how it came to the conclusion which were sent to the client.”
Scrambling for a Way Out
The user admitted that the data appeared to be “complete bullshit” and turned to Reddit for urgent advice. “So, what can we do to save her job?” the user asked.
In response, one Redditor advised damage control with a cover-up explanation: “Say you were using placeholder data and it accidentally got included in the version sent to the client.” Another commenter was more blunt, writing: “Admit the truth, face the consequences, and learn the lesson that ‘ChatGPT can make mistakes. Check important info.’”
A Cautionary Tale About AI in the Workplace
The story highlights the risks of over-reliance on AI tools for professional tasks that demand verifiable accuracy. While ChatGPT can assist with brainstorming and drafting, experts often caution that it can produce errors, misapplied methods, or outright fabricated results if unchecked.