If LLM/Agents contribute to software development, how does the role of software engineers evolve?
SE should focus on :
- System design
- Integrations
- Project management
etc
or the job will disappear in 10 years ?
or the job will disappear in 10 years ?
I'm kidding. Software engineering roles of the "glue things together" kind will evolve into a 'technical product' role, defining how systems work, especially at the edges where systems interact. Architects and staff engineers already do this.
Lower down the food chain engineers who actually invent new things will do the same, only quicker if they have an AI tool assisting with the grunt work of testing, boilerplate, etc.
Engineers who don't really make new things will be replaced by 'low code' style apps that spit out clunky apps. That's already happening though. It's not a change. Those engineers won't leave the industry - they'll get to do more interesting things.
Even at the entry level things won't really change that much. Expectations will go up a bit but people will still have to learn the trade.
Where things will really change is the sheer amount of software we'll have. AI is going to make things go faster, so businesses will be able to automate everything. Dev resource will stop being the constraint that holds businesses back from doing everything they want to do with software. That's a game changer. Society will struggle to keep up.
Sounds like the job of a programmer wouldn't go away very near term in such a scenario. It might well become far less glamorous, less "technical", and not as well paid though. Comparable to an accountant maybe.
As someone who started their career right after the dotcom bubble burst, various people asked me "Why are you doing this? There's no money in software anymore". Turns out there sure was. Back then I didn't know that, and I didn't care. I didn't get into it for the money, I got into it because I had a passion for it. Maybe it'll become more like back then again. Doesn't seem so bad to me.
Possibly, but if SaaS tool providers make APIs so AI tools can consume the data, that would open up a huge ecosystem of AI 'plugins' that would be immensely valuable. I think OpenAI have something like that already.
I can encourage everyone to read good old "Domain Driven Design", there's a world of interesting problems and work out there beyond the purely technical side of things. And I think that part is a lot more difficult to automate than creating a WordPress site from a .psd.
Thank you for putting this into words as it happens to approach my thoughts on it. Our societies are already changing at a crazy pace that makes people not entirely comfortable. Now that this comfort will be lessened still, do you think people will demand changes not completely dissimilar from preserving horse buggys? The pace of change may end up being so fast, that not artificially limiting it may end up pissing a good portion of the planet.
Better look for architecture and management jobs, only a few select ones would still be allowed into the magic tower.
Naturally after a heavy round of letcode interviews that have nothing to do with writing AI algorithms.
But I agree that typical line management jobs will evaporate along with the engineering ones.
It’s a conundrum for “big tech” imo because if they give you too much access to their “AI” then yes, that might pose an existential threat to their core businesses. This is likely the reason for all the regulatory capture hopes and dreams.
Test reviews in the long term (E2E type). Effectively checking the correctness of the system as a whole to fulfill some business process. Exploratory testing as well to find edge cases and gaps.
[0] I write a bit about this here: https://chrlschn.dev/blog/2023/07/interviews-age-of-ai-ditch... and here https://coderev.app/
I think tests are good way of doing that, or some special requirements language will emerge that AI agents can use. I don't think English is particularly good way of describing things exactly. It often has ambiguous meanings which is where the bugs will come from.
In terms of stuff like AutoDev, that's a product person, right?
I think at least for now, we will still need human "customer development."
You can’t possibly think these things will remain clowns forever.
These things aren’t “Markov chains” - the architecture is significantly more scalable, which is exactly why this time is different.
Transformers with finite block size have a finite number of states, so they are Markov chains.