I’m curious as to how others are tackling this issue: How do we get AI users to know that AI is giving them garbage?

For those of us who have years of experience dealing with misinformation, but also a whole lot of experience with actual true facts, there’s a bit of “spidey-sense” that allows you to notice that something seems off. If it’s an area you are well-versed in, you can straight up tell it’s wrong. How do you deal with it when you move into a job/task you aren’t familiar with?

Folks who are junior level workers right now have this the worst because they don’t have a bank of experience to rely on to help with this, but I’ve hit this recently while I’ve been using AI to build out some stuff with Phaser. I don’t know anything about this, and very little front-end building experience in general, so I’m relying on the AI completely to know how to do things.

When things go wrong, I can leverage my experience to debug and get an idea of where the issue might be, but I can’t tell by looking at a response whether the tools are doing the right thing or not. I asked it to modularize some code and it gave a seemingly plausible structure and started filling things out… and left some of the files it created as blank. That was an obvious red flag, but I was then left in the situation of having no idea what it just did with the code and what it removed and lost, or even if it would work at all. (SPOILERS: IT DID NOT).

So how are you tackling this? How do we train up on identifying AI garbage in areas we don’t have experience in?