Change management problem rarely mentioned when pushing AI to engineering teams

I manage two engineering teams and spent most of last year assuming AI tooling adoption would be self-sustaining once I handed people access. You know what happened, it was not. The more interesting problem turned out to be organizational, not technical: how do you build new habits into a team that is already at capacity? Curious whether others found that the bottleneck was individual reluctance, workload, or something structural they had to change at the process level.

Another episode of AI psychiatrist:

- Please make yourself comfortable on this virtual couch and tell me about your problem...

- We've chosen AI for its deep knowledge and universal capability, yet it's sadly incompetent and delusional requiring constant supervision.

- And how is it trained?

- It's trained on all recorded human activity.

- I see...

...The art of effective AI is not in seeking solutions, but incanting the right prompts.

Let's meet again tomorrow.

That will be $50,000 in tokens