When did you learn to trust AI less, or more?

Quick question for the room, because a study out this week put numbers on something I think about a lot.

Researchers followed university faculty through a short, self-paced AI training and measured what changed. The interesting part was not that they used AI more afterward. It was that they got more confident and more skeptical at the same time. They trusted the tool more AND verified it more.

That combination is rare. Most people I meet are stuck on one side. They either avoid AI entirely, or they paste its output in without reading it. The middle, using it boldly while checking it honestly, seems to only come from actually practicing, not from being told the rules.

So here is what I am curious about. When you first started using AI for real work, what was the moment that taught you to trust it less, or more? Was there a specific time it burned you, or a time it genuinely surprised you?

Drop it below. I read every one, and the patterns in these stories are usually better than any study. If you want the backstory, it is in today's post: https://www.harvestkernel.com/faculty-ai-literacy-micro-credential-sdsu-study/