Cluely is a new AI app that listens to your conversations and feeds you the best answers. The team behind it created a phrase designed to be misunderstood and polarizing, but saying that it lets you cheat on any conversation. Cheating like a calculator is cheating a math test, of course.

If you choose to use an app like this, as its founders intend for you to, you’re going to rely on it more and more for the right answer. Maybe you might even get some good results at first. Maybe, because you get fed all the “right” answers, you start outperforming your coworkers and competitors.

The thing is, you didn’t really do it. The AI’s LLM did it. You just happened to be the human agent that spoke the AI’s script. The person you’re giving the answers to isn’t responding to you, they are responding to the AI. You become a part of the AI’s software supply chain, rather than making AI a part of your human supply chain.

“If you’re on a date, and the AI is doing all your talking for you, what do you really have to offer the other person besides a warm body?” my friend Stephen writes.

What a great question. What do you have to offer, that the AI won’t or can’t? There’s always accountability, which someone will trust you with if you can review the AI’s work. There’s also trust and rapport, both of which are more valuable when it’s authentic. (Conversely imagine if you found out that one of your new friends was just reading off an AI script that was responding to your genuine answers. Wouldn’t you feel creeped out?)

Indeed, the right answer is important. However, developing the judgment autonomously to get to the right answer, as contexts change, is just as important. And if an AI is feeding you all your answers, then you can be easily replaced by someone else who can present the AI’s script better or for cheaper.