Ask HN: Future of Programming?
With AI writing most of the code how does the future of programming looks like?
2 comments
Programming specifically? Or software engineering?
I ask because LLMs are currently fairly mediocre engineers, even though they're perfectly adequate programmers (usually). I look back over my career, and they can write all the code I ever wrote, but I think they still couldn't have done all the engineering I did. Apart from anything else, they don't seem to be willing to backtrack as much when barking up the wrong tree.
The question is though, how much does engineering even matter in this future? Does it matter that the AI's code scales badly, when e.g. every restaurant and café can get their own fully customised CRM that does what they want and nothing else, rather than all of them having to buy the same off-the-shelf solution that has to be made to work for every business smaller than Burger King? Likewise, every village being able to vibe-code their own social network just for their residents doesn't need to care if this network has an O(n^2) problem because n ≤ 6000, not 6e9. Do you need all of Photoshop, or do you just want to get rid of the graffiti in background, add your logo, and theme the whole thing for the recent sports result/village fête/holiday/season?
That said, and this goes with the Photoshop example, I find myself questioning if perhaps all the software that needs to exist has already been written. I struggle to think of anything I could go into business to create in this field. Well, except games, but even there game engines already exist.
I ask because LLMs are currently fairly mediocre engineers, even though they're perfectly adequate programmers (usually). I look back over my career, and they can write all the code I ever wrote, but I think they still couldn't have done all the engineering I did. Apart from anything else, they don't seem to be willing to backtrack as much when barking up the wrong tree.
The question is though, how much does engineering even matter in this future? Does it matter that the AI's code scales badly, when e.g. every restaurant and café can get their own fully customised CRM that does what they want and nothing else, rather than all of them having to buy the same off-the-shelf solution that has to be made to work for every business smaller than Burger King? Likewise, every village being able to vibe-code their own social network just for their residents doesn't need to care if this network has an O(n^2) problem because n ≤ 6000, not 6e9. Do you need all of Photoshop, or do you just want to get rid of the graffiti in background, add your logo, and theme the whole thing for the recent sports result/village fête/holiday/season?
That said, and this goes with the Photoshop example, I find myself questioning if perhaps all the software that needs to exist has already been written. I struggle to think of anything I could go into business to create in this field. Well, except games, but even there game engines already exist.
On a three panel glass setup, I'd like to see trees on the left, graphs on the right, the AI hyper explores those trees and graphs, on the center screen I'd like to see three areas of focus, the AI and me hyper explore summative representations, there's an open metaverse teaming aspect, when all three focus rings overlap the problem is solved to an engineering degree that holds water.