Hacker News Reader

Comment by dust42 | Hacker News Reader

> This stuff is relatively new, I don't think anyone has truly figured out how to best approach LLM assisted development yet.

Exactly. But as you say, there are so many people riding the hype wave that it is difficult to come to a sober discussion. LLMs are a new tool that is a quantum leap but they are not a silver bullet for fully autonomous development.

It can be a joy to work with LLMs if you have to write the umpteenth javascript CRUD boilerplate. And it can be deeply frustrating once projects are more complex.

Unfortunately I think benchmaxxing and lm-arena are currently pushing into the wrong direction. But trillions of VC money are at stake and leaning back, digesting, reflecting and discussing things is not an option right now.

View on HN

> But as you say, there are so many people riding the hype wave that it is difficult to come to a sober discussion. LLMs are a new tool that is a quantum leap but they are not a silver bullet for fully autonomous development.

While I agree with the latter, I actually think on former point - that hype is making sober discussion impossible - is actually directionally incorrect. Like a lot of people I speak to privately, I'm making a lot of money directly from software largely written by LLMs (roadmaps compressed from 1-2 years to months since Claude Code was released), but the company has never mentioned LLMs or AI in any marketing, client communications, or public releases. We all very aware that we need to be able to retire before LLMs swamp or obsolete our niche, and don't want to invite competition.

Outside of tech companies, I think this is extremely common.

> It can be a joy to work with LLMs if you have to write the umpteenth javascript CRUD boilerplate.

There is so much latent demand for slightly customised enterprise CRUD apps. An enormous swathe of corporate jobs are humans performing CRUD and task management. Even if LLMs top out here, the economic disruption from this alone is going to be immense.

It is delusional to believe the current frontier models can only write CRUD apps.

I would think someone would have to only write CRUD apps themselves to believe this.

It doesn't matter anyway what a person "believes". If anything, I am having the opposite experience that conversing with people is becoming a bigger and bigger waste of time instead of just talking to Gemini. It is not Gemini that is hallucinating all kinds of nonsense vs the average person. It is the opposite.

There is a critical failure in education - people are not being taught how to debate without degenerating into a dominance debate or merely an echo chamber of talking points. It's a real problem, people literally do not understand that a question is not an opportunity to show off or to dominate, but is a request for an exchange of information.

And that problem is not just between people, this lack of communication skill continues with people's internal self conversations. Many a bully personality is that way because they bully themselves and terrorize themselves.

It's no wonder that people can use AI at all, with how poorly people communicate. So the cluster of nonsense that is all the shallow thinkers directing people down incorrect paths is completely understandable. They are learning by doing, which with any other technology would be fine, but with AI to learn how to use it by using it can serious damage one's cognitive ability, as well as leave a junkyard of failed projects behind.

I blame chats and online forums for banning anything off topic and or political/ religious discussions.

Which led to:

echo chambers

Forgetting how to debate and discuss things.

You needed to go full out with guns blazing to get seen before you vanished and the discussion getting locked or deleted.

I’m not sure you read my comment. I didn’t claim LLMs have reached a ceiling – I’m very bullish on them.

The point I was making is about the baseline capability that even sceptics tend to concede: if LLMs were “only” good at CRUD and task automation (which I don’t think is their ceiling), that alone is already economically and socially transformative.

A huge share of white-collar work is effectively humans doing CRUD and coordination. Compressing or automating that layer will have second- and third-order effects on productivity, labour markets, economics, and politics globally for decades.

even for CRUD I'm finding it quite frustrating. The question is no longer whether AI can write the code you specify: it can

It just writes terrible code I'd never want to maintain. Can I refactor and have it cleaned up by the AI also? Sure... but then I need to specify exactly how it should go about it and eugh should I just be writing this code myself?

It really excels when there are existing conventions within the app it can use as example

Reply on news.ycombinator.com