Hacker News Reader

Comment by proc0 | Hacker News Reader

Just watching the highlights and I think this is mostly correct with some big caveats. The majority of software engineer is everything that happens outside of writing code, from gathering requirements to communicating the changes. Writing the code is like building the physical structure of a house after the materials have been processed, the blueprints have been designed and approved, etc. AI will be the machinery that allows programmers to build things more quickly. At the same time, there are a lot of admin and management tasks that AI will also help with so it's possible we'll see AIs for every role and teams will shrink in size but accomplish the same amount of work (which is unfortunate, but maybe we can start demanding 4 day work weeks?)

View on HN

There is no reason that AI can't be better than humans at requirements elicitation and communication. AI can have necessary domain knowledge about the company and could absorb extensive information provided. It could use analogies to explain concepts at a level comfortable for anyone.

> There is no reason that AI can't be better than humans at requirements elicitation and communication.

Actually there is, one really good reason, even if AI became sentient, it's still a machine, not a human with a soul who can reason and feel passion for a requirement and its solution and implementation.

Yes, I think this will come sooner than AI doing competent level coding (at least for large enterprise codebases). There is a lot of admin level work that is relatively straight forward doing summaries and status updates.

> gathering requirements to communicating the changes.

Right, but those are soft skills. They're important but they're not why SWE's make big bucks. SWE's make big bucks because as he says- it's a lot of hard work to gain the skills to turn an idea into instructions. That's programming.

At the bottom of everything it's just machine instructions. We have all these languages and conventions because humans need abstractions. When models get really good will we won't need programming at all. Tell the model what you want and it generates machine code directly.

And how do you debug it when it doesn’t work and the LLM can’t find the problem?

You go grab the one SWE with an education on the whole team or contract it out.

It's not clear to what extent it's soft skills since gathering requirements often involves pushing back and negotiating what is possible on the technical level while at the same time taking into consideration organization goals and even market considerations. When it does get there, I would expect almost every other role to be completely automated. If you can just tell an AI the requirements and it just figures out the best way to implement it and release it, then I'd say we're very close to a singularity where machines will just take over civilization.

Reply on news.ycombinator.com