In the office of many senior engineers you’ll find a bookcase full of shelf trophies — books meant to signify to the other engineers this person’s taste and knowledge. I see a bunch of old UNIX books, The C Programming Language, The Practice of Programming, or a Smalltalk book? Throw in some modern software design books? This person knows ball. They’re interested in how things work and how they got that way.
Looking at my own books recently got me a little depressed.
Now that (I’m told) AI is writing everyone’s code, what is the value of any of them other than historical curiosity? People are prompting AI to write in languages they don’t know and letting another AI review the output. Does all of our knowledge matter anymore? How does anyone know how to prompt to get the desired result — a maintainable piece of software — if their only experience is prompting and re-prompting until things look right to them?
Is this the last generation of software engineers? I don’t think so. I hope not. If performance sucks, is it better to understand why or just to put the LLM into a loop and tell it to make things faster until it does? Do we trust how it got there?
I keep thinking about what Daniel Jalkut said:
My take on AI is, essentially, everybody who’s against it is too against it and everybody who’s for it is too for it.
I’m not too against AI, because I see its utility and understand how to use it. The people who say it’s a slop machine and still joke they’ll get rich cleaning up AI-generated code are delusional. Why would anyone pay you to fix their AI code when they could spend $1000 in tokens to have AI rewrite the whole thing?
At the same time, I feel that AI has been a net negative for humanity. Everything that was bad before is a bit worse now. Inequities are reinforced. AI is bad for the environment, data centers are forced onto vulnerable communities who don’t want them, and it concentrates power in some truly terrible people (Altman, Musk, Thiel). It’s a weird cognitive dissonance to have, which is probably why I can’t stop thinking about it.
This might be my own cope and lack of imagination, but the way I see things right now is that we are probably going to need people who care and understand how things work. Even if I’m right, who will those people be in the future? Companies already weren’t hiring enough junior engineers. How can a generation raised on “code is a solved problem” even know what to ask for? If you don’t know what an N+1 query is, how do you direct an AI to not do that?
So here I am. Looking at a shelf full of books and wondering if knowledge matters anymore, and if not, what replaces it.
Leave a Reply