When AI Is Not Such a Good Idea

Subtitle

It’s not about what we can outsource but what we want to — and especially what we don’t want to.

Published
November 1, 2026
Author

Irina Rokosz

Remember that viral X post?

It became so popular because it’s true. There are certain activities you want to keep doing yourself: playing music, attending live concerts, meeting with friends, writing and reading books, snowboarding, painting pictures, camping, pottery, even ironing your clothes if, during this time, you call a friend and talk for hours.

I remember when I was going to dancing classes. My mom asked me when I’d finally learn how to dance and stop going altogether. I couldn’t help laughing, even though she was dead serious.

Not everything we do is for the result: many things we do are just for fun, because we actually love the process. The result is simply the end of one iteration and the beginning of another. We also don’t get paid for many of these activities (hence, we call them hobbies) — in fact, quite often we are the ones paying.

But more importantly, we feel a certain pride in a job well done. A level of satisfaction from the craftsmanship we put into our work (whether paid or unpaid, it doesn’t matter). And we become a bit better in the process of mastering our craft.

Which brings me to this story: Recently, I attended a conference where there were a lot of AI talks. (Is there a conference nowadays where there are no AI talks?!) Managers were enthusiastically explaining how many lines of code AI can write in no time, and the software engineers will only get to review the results afterwards. All went well until a software engineer spoke up and said, “I’d like to flip this one on its head. What I like the most is writing code. And what I hate the most is reviewing code. So maybe I should be the one writing the code, and AI reviewing my code, not the other way around. Just food for thought.”

The silence after he spoke was palpable. Clearly, no one has considered this perspective or asked the software engineers what they loved and hated doing. Everyone was too focused on the efficiencies AI could bring. Have you considered if these efficiencies are actually worth implementing? What do your people love and hate doing — have you asked them? Would it even matter?