It also assumes the people will somehow develop the skill to verify whether the LLM-result is "good". That's not guaranteed to happen and it's not a new proble

It also assumes the people will somehow develop the skill to verify whether the LLM-result is "good". That's not guaranteed to happen and it's not a new problem, since it happens when managers are unable to detect problems in the team's product or output.