I keep hearing people claim that it is no longer possible to review all the code AI produces. There is simply too much of it. At first, it was a few people on YouTube and Twitter.
Recently, I heard two law professors make the same argument about human involvement in AI decisions: AI works so quickly that humans cannot review every decision.
I get stuck on the word “impossible.” It doesn’t feel right.
We could slow down. We could use more resources. We could process fewer decisions. That may be impractical, but it is not impossible.
I don’t want to be a stickler for words. But we are making a choice. We can make it explicit and decide that the time, cost or effort of human review is not worth it. Or we claim that it’s not possible anymore and act like we have no choice.