I raised this as a concern at the corporate role I work in when an AI tool that was being distributed and encouraged for usage showed two hallucinated data points that were cited in a large group setting. I happened to know my area well, the data was not just marginally wrong but way off, and I was able to quickly check the figures. I corrected it in the room after verifying on my laptop and the reaction in the room was sort of a harmless whoops. The rest of the presentation continued without a seeming acknowledgement that the rest of the figures should be checked.

When I approached the head of the team that constructed the tool after the meeting and shared the inaccuracies and my concerns, he told me that he’d rather have more data fluency through the ease of the tool and that inaccuracies were acceptable because of the convenience and widespread usage.

I suspect stories like this are happening across my industry. Meanwhile, the company put out a press release about our AI efforts (literally using Gemini’s Gem tool and custom ChatGPTs seeded with Google Drive) as something investors should be very excited about.