What do you mean? I've been productively shouting at LLMs (particularly via ChatGPT's advanced voice mode) for quite a while now, and they're reasonably receptive to my shouting. More recently , I've been (textually) shouting at OpenaAI's async Codex agents, and they're quite receptive too.
We're obviously not quite there yet, but I don't see any inherent limitation to human managers' ability to shout at AI agents. I'll go even a step further and say that knowing that the AI doesn't actually have feelings that I can affect, and that it has limited context helps make my shouting more productive, focusing on the task at hand, rather than on its general faults, as I might when I'm angry at a human.
We're obviously not quite there yet, but I don't see any inherent limitation to human managers' ability to shout at AI agents. I'll go even a step further and say that knowing that the AI doesn't actually have feelings that I can affect, and that it has limited context helps make my shouting more productive, focusing on the task at hand, rather than on its general faults, as I might when I'm angry at a human.