>> The result will be that mediocrity takes over because the AI will be in charge of the boilerplate, which is often the foundation for the rest of your code.
Your take is interesting, but I'll provide an alternate point of view from my experience.
I have spent my career working in a low-code environment. There's a boiler plate framework into which people add code.
Since most of all programs are similar (data in, data stored, data processed, data out etc) this allows for high productivity because almost all time is spent on things that are unique to the domain space, not programming the common stuff. It means individual programmers, and very small teams, have written and maintain huge systems.
But what's interesting to me is the boilerplate code. My career has been spent improving the underlying boilerplate, and extending it to new feature areas. This turns out to be a great lever. Improve the boiler-plate and _existing_ programs get better, with minimal effort. Sometimes no effort.
It's kinda like the way I get TLS 1.3 support simply by upgrading my OpenSSL libraries. But at a source level. Which is then retro-applied to existing source code.
Instead of being mediocrity the goal is to always keep improving the boilerplate because that in turn improves a lot of programs.
I say this not to tell you nothing will change. It will change. But change in programming is fun. It stretches the mind, and opens up new horizons. The AI can write code, but that just becomes the next tool, like IDEs were, like code complete, online help or Stackoverflow. I didn't have any of those things growing up, but their introductions allow me to dream bigger, be better, make more.
If you look at it from the bottom, then yes. SAP/Salesforce/IBM DB2/Oracle/OpenERP are ways for company departments to all work on one common set of data.
Your take is interesting, but I'll provide an alternate point of view from my experience.
I have spent my career working in a low-code environment. There's a boiler plate framework into which people add code.
Since most of all programs are similar (data in, data stored, data processed, data out etc) this allows for high productivity because almost all time is spent on things that are unique to the domain space, not programming the common stuff. It means individual programmers, and very small teams, have written and maintain huge systems.
But what's interesting to me is the boilerplate code. My career has been spent improving the underlying boilerplate, and extending it to new feature areas. This turns out to be a great lever. Improve the boiler-plate and _existing_ programs get better, with minimal effort. Sometimes no effort.
It's kinda like the way I get TLS 1.3 support simply by upgrading my OpenSSL libraries. But at a source level. Which is then retro-applied to existing source code.
Instead of being mediocrity the goal is to always keep improving the boilerplate because that in turn improves a lot of programs.
I say this not to tell you nothing will change. It will change. But change in programming is fun. It stretches the mind, and opens up new horizons. The AI can write code, but that just becomes the next tool, like IDEs were, like code complete, online help or Stackoverflow. I didn't have any of those things growing up, but their introductions allow me to dream bigger, be better, make more.
AI is an opportunity for those with eyes to see.