To create things via AI without being able to comprehend the output is to trust completely in the agent(s). Operating that way is more optimistic than experience has taught me to be.