This is an overstatement. There are still humans in the loop to do the prompt, apply the patch, verify, write tests, and commit. We're not even at intern-level autonomy here.
Plugging DeepSeek R1 into a harness that can apply the changes, compile them, run the tests and loop to solve any bugs isn't hard. People are already plugging it into existing systems like Aider that can run those kinds of operations.
Absolutely the AI. At that point in the future I'm presuming that if something breaks it's because an external API or whatever dependency broke, not because the AI code has an inherent bug.
But if it does it could still fix it.
And you won't have to tell it anything, alerts will be sent if a test fails and it will fix it directly.
I'm very sorry, but the goalposts are moving so far ahead now, that's it's very hard to keep track of. 6 months ago the same comments were saying "AI generated code is complete garbage is useless, and I have to rewrite everything all the time anyways". Now we're onto "need to prompt, apply patch, verify" and etc.
Come on guys, time to look at it a bit objectively, and decide where we're going with it.
Couldn't agree more. Every time these systems get better, there are dozens of comments to the effect of "ya but...[insert something ai isn't great at yet]".
It's a bit maddening to see this happening on a forum full of tech-literate folks.
Ultimately, I think to stay relevant in software development, we are going to have accept that our role in the process could evolve to humans essentially never writing code. Take that one step further and humans may not even be reviewing code.
I am not sure if accepting that is enough to guarantee job security. But I am fairly sure that those who do accept this eventuality will be more relevant for longer than those who prefer to hide behind their "I'm irreplaceable because I'm human" attitude.
If your first instinct is to pick these systems apart and look for things that they aren't doing perfectly, then you aren't seeing the big picture.
Regarding job security, in maybe 10 years (human and companies are slow to adapt), I think this revolution will force us to choose between mostly 2 career paths:
- The product engineer: highly if not completely AI driven. The human supervises it by writing specification and making sure the outcome is correct. A domain expert fluent in AI guidance.
- The tech expert: Maintain and develop systems that can't legally be developed by AI. Will have to stay very sharp and master it's craft. Adopting AI for them won't help in this career path.
If the demand for new products continue to rise, most of us will be in the first category. I think choosing one of these branch early will define whether you will be employed.
That's how I see it. I wish I can stay in the second group.
> - The product engineer: highly if not completely AI driven. The human supervises it by writing specification and making sure the outcome is correct. A domain expert fluent in AI guidance.
If AI continues to improve - what would be the reason a human is needed to verify the correct outcome? If you consider that these things will surpass our ability, then adding a human into the loop would lead to less "correct" outcomes.
> - The tech expert: Maintain and develop systems that can't legally be developed by AI. Will have to stay very sharp and master it's craft. Adopting AI for them won't help in this career path.
This one makes some sense to me but I am not hopeful. Our current suite of models only exist because the creators ignored the law (copyright specifically). I can't imagine they will stop there unless we see significant government intervention.
Quite the contrary, really. We've been seeing "success stories" with AI translating function calls for years now, it just doesn't get any attention or make any headlines because it's so simple. SIMD optimization is pretty much the lowest-hanging fruit of modern computation; a middle schooler could write working SIMD code if they understood the problem.
There's certainly a bit of irony in the PR, but the code itself is not complex enough to warrant any further hysteria. If you've written SIMD by hand you're probably well familiar with the fact that it's more drudgery than thought work.
It's been probably about 15 years since I've touched that, so I genuinely have no recollection of SIMD coding. But literally, that's the purpose of higher level automation? Like I don't know/remember it, I ask it to do stuff, it does, and the output is good enough. That's how a good chunk of companies operate - you get general idea of what to do, you write the code, then eventually it makes it to production.
As we patch the holes in the AI-code delivery pipeline, those human-involved issues will be resolved as well. Slowly, painfully, but it's just a matter of time at this point?
I mean currently yes, but writing a test/patch/benchmark loop, maybe with a seperate AI that generates the requests to the coder agent loop, should be doable to have the AI continually attempt to improve itself, its just no ones built the loop yet to my knowledge
This is an overstatement. There are still humans in the loop to do the prompt, apply the patch, verify, write tests, and commit. We're not even at intern-level autonomy here.