I practiced this a long time ago when the food if email became too much. I just don't read most of it. If it's important, someone will speak to me about it. This has worked for me ~10 years (I do skim my email, but not often).
Find the minimal number of critical things you feel you should be working on, and work on them. Use what resources are practical to use and ignore the rest. I've got a relatively complex problem to figure out over the next week or two, and one of the developers has been asking Claude and posting much of the output to Confluence pages, but even he says it's too much information to process and, as a result of a few things we've learned along the way so far, much of the AI summarized information is now out of date.
I just try to concentrate on doing a good job with the information I am able to process.
I think unfortunately this is a stage people have to go through when they're first exposed to AI. You just have to let it play out for them to realize actually they still need to cooperate with humans to get anything meaningful done.
First rule is - if you don't write it, don't expect me to read it. If you AI-write something, then let me get an AI to read it too.
The only solution I can think of is to get AI implemented upstream in the planning phases so that it results in specs for spec-driven development. Your boss will be able to directly contribute without bothering anyone, but he will also think developers are obsolete. You need to warn him about the dangers of vibe-coded shit, but either way there's no positive scenario these days.
you should take the information the same way whether it was produced by him manually or with the help of AI. If you won't someone else who needs a job more will.
I practiced this a long time ago when the food if email became too much. I just don't read most of it. If it's important, someone will speak to me about it. This has worked for me ~10 years (I do skim my email, but not often).
Find the minimal number of critical things you feel you should be working on, and work on them. Use what resources are practical to use and ignore the rest. I've got a relatively complex problem to figure out over the next week or two, and one of the developers has been asking Claude and posting much of the output to Confluence pages, but even he says it's too much information to process and, as a result of a few things we've learned along the way so far, much of the AI summarized information is now out of date.
I just try to concentrate on doing a good job with the information I am able to process.
I think unfortunately this is a stage people have to go through when they're first exposed to AI. You just have to let it play out for them to realize actually they still need to cooperate with humans to get anything meaningful done.
First rule is - if you don't write it, don't expect me to read it. If you AI-write something, then let me get an AI to read it too.
The only solution I can think of is to get AI implemented upstream in the planning phases so that it results in specs for spec-driven development. Your boss will be able to directly contribute without bothering anyone, but he will also think developers are obsolete. You need to warn him about the dangers of vibe-coded shit, but either way there's no positive scenario these days.
I don't have any advice, but I sure do feel bad for you!
Just ignore it and they'll get bored and move on to the next thing
you should take the information the same way whether it was produced by him manually or with the help of AI. If you won't someone else who needs a job more will.
Have you talked to him about it?
And what do your colleagues feel?
I feed it back to AI to respond with em-dashes, en-dashes and clearly Claude-like writing.