For the most part AI’s only know how to do what we know how to do, but it may be able to do it better.
It can be argued that certain learning systems such as Alpha GO can play GO in ways better than we know how to do. That’s true for certain types of problems.
AI can’t manage humans better than we know how to manage humans, but they can, for example, be infinitely patient, which can be a useful management trait that humans generally lack.
It also very much depends upon what you mean by manage.
The Uber system for example is directing thousands of drivers to locations, meetings, perform deliveries, etc. In a sense that is directing / managing a whole lot of human resources. It is selecting routes, prioritising pick ups, selecting optimal use of resources, including human resources, to achieve required goals.
Likewise SAP systems are planning production runs, managing inventories, giving instructions to humans, scheduling leave and close downs for maintenance, etc. etc.
In one sense this hardly amounts to high level management skills.
In another sense, a lot of people management really is this basic; organise people, tell them what to do, direct them to perform particular tasks and make sure goal objectives are met. Yes - both Uber and SAP are doing these things.
But where do you draw the line?
A traffic light also tells drivers when to stop and when to go. Is that people management? Is it AI?
Well, it can be if it is part of an entire traffic network control system for a city. But one set of traffic lights operating automatically on its own really isn’t AI even under my most basic definition (responding to information about changed circumstances).
A light turns on when you click the switch. That isn’t AI.
The light turns on when a sensor advises that it is now dark.
?
As with so many things relating to AI, it really depends upon your definitions.
AI forces us to re-examine a lot of definitions.
Hope this helps.