The model handed me a working solution.
It ran. The tests passed.
And I realized I couldn’t explain it.
That scared me more than a bug would have.
Because the day it broke,
I’d be standing in front of code I didn’t understand.
So I made a rule for myself.
Let it write, yes.
But I don’t ship what I can’t explain.
I read it. I trace it. I make it mine.
The model can produce the answer.
It can’t do the understanding for me.
If I can’t explain it,
I don’t really own it —
I’m just borrowing it until it breaks.
– Serguey Asael Shinder
Leave a comment