The article makes some salient points. The Jevons paradox makes sense, assuming AI does make software development more efficient. From my own experience, AI does bring a modest productivity boost when used correctly, though it takes a competent and experienced engineer to know how to create the correct feedback loops and properly review the generated code in order to create something that isn’t AI slop.
I also agree that the current job market isn’t due to AI, but instead other factors like irrationally exuberant tech hiring during the pandemic and high interest rates.
Personally, I consider AI to be a permanent part of my tool belt—something I don’t mind using when it’s the right tool for the job (and it isn’t always). To the author’s point, though, the current tech may in fact not be sustainable cost-wise. If employers actually had to pay the real price for the tooling that reflects all of trillions that went into to developing it, the costs of the data centers, etc., then it may cost so much to use it that the modest productivity gains won’t justify the price.
