The question is ultimately if you would have been better served with doing your own investigation rather than relying on AI, did you retain the knowledge, and was the information complete/accurate.
I use AI, and it helps me navigate a code base far faster than without. Like if I need to know if a function is used, the AI could tell me in a minute, or I could be using grep on the code base for a half hour.
I know to not trust it and to verify the output. It crawled deep into a complex submodule I didn't understand and helped me devise a fix. I didn't check that fix in as it had a few errors code-wise that would have become a maintenance nightmare and a lingering bug if someone added a member variable to a struct.
Now, I could've gone through the code myself, and probably taken a few weeks to properly understand it. But AI did it, and I had it produce a few documents on the design of the system. Those documents let me understand what's going on so I can see how it's doing things and how I can actually see the code better. Granted, with a few false starts.
I would say the AI helped me - mostly because I was told to treat it as a co-worker than as an expert to ask to do something. So I would guide the AI into what needed to be done, and then leave the final details for the AI to do it. Then I would review what the AI did
My AI token spend isn't huge compared to other users, but that's because I'm judicious with how I use it. It spends more time doing cheap things before coming out with a result that's relatively succinct and correct so it's not wasted on expensive output tokens.
Everything I used AI for, I could've done it myself. It just means I have an assistant who saved me likely a month of work trying to understand the code and another month devising a fix and I managed to get it done is about 2 weeks and in a way I could understand what it did, and the code reviews of the changes make a lot of sense.
I still remain a skeptic and find these highly guided uses make me more efficient while letting me not get too bogged down in the details. I constantly question whether I could do it myself or if I'm relying on the AI and make sure the AI output makes sense. (I don't trust it to commit code to git, for example, so I have it make all the changes it wants, then I test them and review them before submitting).