Have you heard of The Mythical Man-Month? It’s supposedly1 a must-read for managers. The main thesis: adding manpower to a late software project makes it later.
Similarly, I think replacing humans with AI in a late software project also makes it later.
Let’s say you have 10 engineers. One is an expert in security. One is good at UI. One runs the servers. One makes mobile apps. One knows how the payment system work. Each person has domain knowledge that took years to build.
Then you fire the security guy and expect the other nine to use AI to “fill the gap”. What could go wrong?
This is a simplified example. You can extrapolate it to bigger companies. A bigger company has multiple products. Each product has multiple components. Each component requires multiple domain experts. Maybe you need some redundancy because otherwise the security guy can’t take a day off.
But the tech industry is doing this right now. They overhired during the zero-interest-rate era and they’re correcting. They all say it’s because of AI. It sounds better than “we hired too many people when money was free and now we need to cut costs”.
So they fire people, give the remaining engineers an AI tool, and call it a productivity gain. It works until the server goes down at 2am and you realize you fired the only person who knew how to fix it.
And from the outside, we won’t see news stories like this because no company is going to say “we replaced our engineers with AI and it went badly”. They’ll just say “we experienced an incident” and quietly rehire.

-
Because everybody quotes it and few people follow the advice. ↩