Here is a razor for the age of AI.
When building the strongest case against your own position costs almost nothing, not having one says something about your honesty, not about the strength of your position.
It used to be expensive to argue against yourself. You had to find someone smart who disagreed, get them to care, and sit through it. Most of us skipped that step and called it conviction. Now the opposing case is a paragraph away. Paste in what you believe, ask for the best argument against it, and read it. Two minutes.
So the excuse is gone. If I hold a position and I have never asked for the strongest case against it, that is a choice I made, not a limit I hit.
A few implications I am still grappling with.
First, “I hadn’t thought about that” gets harder to say with a straight face. Second, the razor cuts toward me first. My positions on hiring, on parenting, on what my next job should be: I have run very few of them through this. Third, a machine’s objection is not the same as a person’s. The model gives me the argument. It does not give me the discomfort of someone I respect saying it to my face, and the discomfort might be the part that works.
That last one is the open question. Is the razor complete with the machine alone, or does real human pushback have to be part of it? I do not know yet.
The ask: pick one thing you believe and ask for the best case against it. Then notice whether you actually read it.