If you've worked in the AI field long enough -- and this includes ML/AI modeling as well as genAI/LLMs -- you, too, have experienced the model that is "confidently wrong."

(A colleague refers to these "mansplainer" models. I won't reveal their name. But if they're reading this: you know who you are. And you're awesome.)

This LinkedIn post by David Villalon, about a genAI-based error in a financial spreadsheet, drives the point home.

The take-away lesson? Same as always: you have to check a model's work. And the more important the outcome, the more you have to check.