This is a useful framing. Treating AI as an async collaborator instead of an instant oracle feels more realistic, especially as reasoning modes get slower. Good workflows should assume latency and help humans stay productive while the model is thinking, not just show a spinner.

I have two or three instances of Claude running at the same time. I can't keep up. It's usually waiting for me, not the other way around.

call the community mental health support team and get a psychiatric assessment/diagnoses.

projecting human conditions (thinking) onto a machine is the beginnings of severe psychotic mental decline.