It's not reasonable to suggest that AI is only going to repeat older patterns that have been trodden before, or 'things that don't matter'.
AI will be writing most new code, by far.
Without even getting into complicated arguments about 'creativity' - the AI is an encyclopedia of best practices, and can think a couple of steps ahead for most things you'll ever want to do.
Like pro chess players thinking they're going to beat the algo with some kind of fancy human creativity.
Developers roles are changing, very fundamentally, you're now 1/2 a layer of abstraction above the code, and you're not going to writing it better than AI (in most cases) any more than a human will be better at sawing wood than the power tools. And yet, carpenters still exist.
The key to this argument is that we won’t need to rely on Anthropic/OpenAI soon — will they exist in the same way they do today in 12-18 months? The “open” models are getting better and better, and people are figuring out ways to make inference run on lesser hardware. It already might be viable for people that don’t expect “instantaneous” and are doing more hybrid development.
But you’re also never going to convince the people who still only run vi on the Linux console, without Xorg…
I don't remember automobiles actively harming your cognitive abilities, nor the cost increasing 5 orders of magnitude. Yet the things actively slow me down in my work by the shear impact of my coworkers using them and having to correct the many mistakes that are made. If your job was producing mediocrity, then yes, AI is awesome. Sorry for the mirror.
Edit: I have my popcorn ready for when the VC subsidies end.
Your job as a developer is not to produce brilliant code - it's to produce code that is so standard, disciplined, off-the-shelf and normative that it's utterly boring.
That is exactly what the AI is good at - 'the boring standard'.
-> AI is not harming anyone's cognitive abilities.
-> Cost of AI is marginal - $200/mo for a tool while devs earn >>$100K, is the cost of tires on a car
-> Token prices are dropping by 90% per year, today's AI is already pretty good, it will be a commodity soon, but we'll continue to pay probably even more than $200/mo because the value is there.
I taught myself BASIC in the 1980's and there's code I wrote in the 1990's that is literally still in production to this day. There's probably a decent chance that I've had code in production since before you were born. Over the span of those decades, it's very clear that there's a fundamental transformation, which is totally unassailable.
It's not reasonable to suggest that AI is only going to repeat older patterns that have been trodden before, or 'things that don't matter'.
AI will be writing most new code, by far.
Without even getting into complicated arguments about 'creativity' - the AI is an encyclopedia of best practices, and can think a couple of steps ahead for most things you'll ever want to do.
Like pro chess players thinking they're going to beat the algo with some kind of fancy human creativity.
Developers roles are changing, very fundamentally, you're now 1/2 a layer of abstraction above the code, and you're not going to writing it better than AI (in most cases) any more than a human will be better at sawing wood than the power tools. And yet, carpenters still exist.