AI tested on people

Not sure if my thread should be here in testing or it concerns philosophy... Actually, it is about both.
Some software devs were put under control of AI, it played a role of their project manager. AI fulfiled its obligations with flying colors but it was a complete fail for humans as a big part of a team was fired by a machine. How We Put Our Team Under Artificial Intelligence Control - Qubit Labs Can we call such testing successful?
IMHO, yes. People will have to suffer a bit to give AI a chance to study. In the nearest future developers will improve AI and it'd be more human-like in terms of emotions.

There is a subforum about cobots, it seems that this thread should be moved there.
The truth is that in this subforum it is a question of robots, physical devices that are made in such a way as to do work with a person and not to cause people physical damage.
But non-material artificial intelligence must also have the ability to work with people and not impose moral damage on people.

By analogy with cobots, we can suggest the term coAIs.

There can be AIs working in unmanned areas, and there can be AIs working with people - coAIs.