Hacker News

points by drstewart 3 days ago | 0 comments

Disagree. The only part an AI can't do is the people aligning and box drawing. The implementation part is trivial. That's who will get replaced.

[–] llmslave 3 days ago

arguably the meeting overhead is actually just people trading domain knowledge. an ai agent can go read their whole code base instead of having that meeting. its not clear how all this shakes out

[–] tripledry 3 days ago

I would say in addition to domain knowledge, it's also about intent. Code does not capture intent, it usually doesn't include the "why" just the "how".

I argue you still need communication for the "why" to get to the "how", otherwise you will just get the wrong thing.

[–] wallst07 2 days ago

Meta: This is a topic I'm very interested in reading about. Does anyone know where people can talk about how AI is changing how people in large orgs work.

I don't mean in a forum where everyone bashes AI, but actually discusses how to navigate this?

It's not here at HN.

[–] lionkor 2 days ago

I'd also be interested!

The AI bashing is a response to the "AI is superintelligence and can write a kernel that's better than Linux" mentality, that is tiring, reeks of gross incompetence, and is almost maliciously wrong.

SOTA LLMs can't reliably implement a single, small feature without producing overly complex code. They don't do engineering.

My approach is two-fold: 1. frontload every conversation with a lot of guidance on how to approach things "like an engineer" (those terms are never uttered, but the guidance basically encodes my workflow as an engineer), to focus on only the problem at hand today, not overbuild things, etc, and 2. not trust a single line of code it produces.

[–] wallst07 1 day ago

I'd be ok discussing with people NOT using AI at work as long as its not built around bashing AI. There are people I work with who still write code, I'm not going to downplay them.

It's mostly around navigating these nuances at work. Even what words to use "I built this" vs "LLM built this"

I know how to design large multi PR projects with multiple agents, but the ORG/LLM navigation , you can't learn by talking to AI.

[–] IcyWindows 2 days ago

Let me know if you find it, I'm also looking

[–] wallst07 1 day ago

I think it's too nuanced of conversation topic to have a dedicated space.

All spaces turn into echo chambers with enough people. I could be wrong.

[–] iugtmkbdfil834 2 days ago

Eh. I haven't seen one. The topic pops up every so often in individual threads, but, to your point, it tends to devolve into AI booster/doomer tug of war. One would think HN would be a natural spot for it, but maybe the easy profile generation prevents it. Maybe there is needs to be a forum with some weird level experience verification?

[–] wallst07 1 day ago

HN falls under the same cause/effect rise/fall of any other large community.

The larger it gets, the more fringe/outlier groups start to take over.

[–] iugtmkbdfil834 2 days ago

In a weird way, I kinda want to encourage that world view. Depending on the definition of 'aligning and box drawing', I am sure it may well be accurate in smaller shops, but I don't know if any company would let AI actually implement stuff. But.. I am ok with people trying. Should be fun to watch; ideally from some distance.

[–] hintymad 3 days ago

I assume that the demand will change. Say, a new generation of companies to replace the old. Besides, those who do have practical experience on building AI-native applications or building AI infra may be in higher demand as well. That is, maybe our experience will become not as valuable as before, therefore we'd need to refresh our experience with a different pace.