Live data from Hacker News

The AI arms race is going to destroy software engineering forever

news.ycombinator.com

11–15 of 15 posts

Re: The AI arms race is going to destroy software engineering forever

#11

Are you a software developer? There will be a huge demand for fixing the bugs that the AI will create. Also, what happen if the AI generated system need a new feature, who will create it, the AI?

So the worst part of the job will become the entire job? Great!

Re: The AI arms race is going to destroy software engineering forever

#12
Since current AI only generates code that has some similarity to code it's seen before, I suspect this will only be an issue for people doing low code or making relatively simple CRUD applications and the like, at least for the foreseeable future. People working on software that requires a lot of domain knowledge won't lose their jobs. Salaries could go down if the demand for software engineers drops, but if the number of new engineers also falls then salaries could stay around the same level, and even increase for niche roles. In the long term we might get into a situation where a small number of engineers are trained to provide training data for the AI, which would be interesting. All of this ks of course pure speculation.

Re: The AI arms race is going to destroy software engineering forever

#13
post #2

But what’s the downside

None except our (software engineers) privileged position not being so privileged anymore

It's really a missed opportunity. In medieval times, we would have created a Programmers' Guild, kept the AI for ourselves, made up some reason AI can't generate code, automated everyone else out of a job, and been the masters of society. Unfortunately we are too altruistic (some would say naïve) these days and let the suits control the AI. Now people are saying we'll be the first to lose our jobs.

Re: The AI arms race is going to destroy software engineering forever

#15
I see your concerns, and I find them very valid, am I am happy that things are this way. I think you are right about some things, but probably very wrong about the outcome here. Software development is about to enter a philosophical stage, on the idea of what it means to be a software developer at all. Even if Prompting becomes close to perfect, or in between perfect and what we have now; there is still going to be a need for people to build/prompt something. There are going to be black boxes, but very unlikely for everything.

I think that a good software developer who knows what he is doing, will look very different from a "software shop" stitching together prompts. Shops doing this, as in the past, will be the low hanging fruits, and at some time will stop being valuable due to sole developers, knowing what they are doing, rather than throwing stuff together.

In my opinion, this type of sole software developer, does now know how to engineer with prompts, take care of deployments, tie into business needs, analytics and sales, within the same (or even less) time it would take him to develop an app before LLM's or AI. This would not be a status-quo, but I would assume, a lot of developers, who are good in their field now, could get their in the future with AI & LLM's.

Now imagine a company staffed with those developers, it would be a completely different playing field, but not so much different from today's world, were developers to date have been dealing with ever increasing complexity, building bigger and larger stuff.

Post reply on HN