You're probably not wrong, but a 99% sensible plan can still be catastrophically bad. The complex knot of decisions to be made in the 1% may completely invalida

You're probably not wrong, but a 99% sensible plan can still be catastrophically bad. The complex knot of decisions to be made in the 1% may completely invalidate whatever the AI already told the higher-ups. This happens even now when humans painstakingly plan something out (usually because someone is abusing their "soft skills" to buy time or keep others from worrying). I think humans are still miles ahead in being able to spot a lie that will derail a project. If you force people to work under an equally incompetent AI middle manager, prepare to see them quit in a mass exodus for somewhere that doesn't. Nobody (except naive college grads) is going to risk getting fired or having their career ruined by an over-glorified cost-cutting measure that prioritizes the convenience of the rest of the business over the quality of the work being done.