Reward people for finding what AI got wrong

Most teams naturally praise the person who found a clever new use for AI, but rarely praise the person who caught it making a mistake, even though both matter equally. I make a point of calling out the second kind of win myself, since I've noticed it's the one that quietly stops happening once it goes unnoticed. Make a point of thanking someone publicly when they catch a bad output before it caused a problem, the same way you'd thank someone for a good idea. This tells the whole team that careful checking is valued work, not an annoying extra step that slows things down. Without that recognition, people quietly stop double-checking because it feels invisible and thankless compared to finding a shiny new use case. A team that feels appreciated for catching mistakes will keep doing it, and that habit is what keeps a tool trustworthy long after the excitement of launch has faded.

Try this today: the next time someone on your team catches a wrong AI answer before it went out, thank them in front of the group, not just privately.