Building trivial tools with AI doesn't mean that devs won't have jobs. It might mean that people won't be able to make a living whipping out brochure-ware web sites for their friends, but that market was just about dead anyway. And maybe it will even be able to handle basic CRUD apps. That doesn't mean the industry is dead, it just means the bar for low-hanging fruit is now hanging a little higher.
I decided to try to do it with AI developers. Tried bolt.new, v0 and then replit. Replit did the work the best. It took me less than half an hour to make the tool I wanted. Now I wanna spend the next 10 days making all these tools myself.
Anyone doing the same?
I believe every Jr dev should be learning how to use code gen tools as they are likely here to stay, but not overly rely on them, and use them like any other tool.
How do you get medium/senior devs in the future?
That pipeline is needed.
or it might also be the opposite and junior devs may be using AI and learn faster.
In my own biz, I stopped hiring junior devs because it's faster for me to use AI for coding than to explain what I need and then validate the implementation with code-review...
Encouraging AI to replace people is just going to make it worse, on top of all the other possible issues.
A not-too-dreary idea: Perhaps the the beginning of the career-arc will change in terms of how quickly developers need to focus on code-reviews and diagnostics (of LLM output) at the expense of "read the docs and follow a guide." Of course that depends on a cultural understanding that LLM output must be checked, and that may take some major industry to become accepted by non-developer bosses, if ever.
I’ve already seen higher level devs ask AIs questions and just copy/paste the answer without checking it as if that’s useful or trustworthy. How is a junior supposed to learn?