Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

LLM Inevitabilism

tomrenner.com

Post reply on HN
Print view View original

161–170 of 1001 posts

Re: LLM Inevitabilism

#161

Post by isoprophlex »

ccortes wrote: post #145

Earlier today I was scrolling at the “work at a startup” posts. Seems like everyone is doing LLM stuff. We are back at the “uber for X” but now it is “ChatGPT for X”. I get it, but I’ve never felt more uninspired looking at what yc startups are working on today. For the first time they all feel incredibly generic

Finally we've managed to disintermediate everything. Even the cutting out of middlemen can now be automated.

A machine stamping out cookiecutter saas businesses. Business model: Uber for "Uber for x".

Who wants to start a goat farming co-op?

Re: LLM Inevitabilism

#162

Post by bradly »

westoncb wrote: post #152
bradly wrote: post #121

Earlier quoted context omitted.

This is what that same GPT4 told me today after trying to get a simple mqttwarn config: ChatGPT: Let’s cut through the mqttwarn quirks and get you a single, complete warn.ini that I’ve actually tested with mqttwarn 0.32 Me: that "I’ve actually tested with mqttwarn 0.32:" How did you perform this test? ChatGPT: I set up a quick local sandbox on my laptop Me: What brand is your laptop? What store did you buy it at? Cha…

There is a skill to it. You can get lucky as a beginner but if you want consistent success you gotta learn the ropes (strengths, weaknesses, failure modes etc). A quick way of getting seriously improved results though: if you are literally using GPT-4 as you mention—that is an ancient model! Parent comment says GPT-4.1 (yes openai is unimaginably horrible at naming but that ".1" isn't a minor version increment). And…

Specific model I was using was o4-mini-high which the drop-down model selector describes as "Great at coding and visual reasoning".

Re: LLM Inevitabilism

#163

Post by mumbisChungo »

imiric wrote: post #112
Animats wrote: post #46

There may be an "LLM Winter" as people discover that LLMs can't be trusted to do anything. Look for frantic efforts by companies to offload responsibility for LLM mistakes onto consumers. We've got to have something that has solid "I don't know" and "I don't know how to do this" outputs. We're starting to see reports of LLM usage having negative value for programmers, even though they think it's helping. Too much eff…

> Look for frantic efforts by companies to offload responsibility for LLM mistakes onto consumers. Not just by companies. We see this from enthusiastic consumers as well, on this very forum. Or it might just be astroturfing, it's hard to tell. The mantra is that in order to extract value from LLMs, the user must have a certain level of knowledge and skill of how to use them. "Prompt engineering", now reframed as "con…

The more I learn about prompt engineering the more complex it seems to be, but perhaps I'm an idiot.

Re: LLM Inevitabilism

#164

Post by laserlight »

mg wrote: post #70

In the 90s a friend told me about the internet. And that he knows someone who is in a university and has access to it and can show us. An hour later, we were sitting in front of a computer in that university and watched his friend surfing the web. Clicking on links, receiving pages of text. Faster than one could read. In a nice layout. Even with images. And links to other pages. We were shocked. No printing, no shipp…

Now that we are sharing anecdotes, here's mine. I asked Cursor to implement a very basic thing in Pydantic, in which I lacked any experience. Cursor spitted out what seemed like a mess to me. After many back-and-forths and cross-checking with documentation, I couldn't make it do it the way I thought it should be. I went ahead and studied Pydantic's well-written documentation. Done. Hours of time saved.

Re: LLM Inevitabilism

#165

Post by godelski »

viccis wrote: post #135
godelski wrote: post #36

Earlier quoted context omitted.

It's insane too. Because many of us working on AI were working on it for different reasons. To me, it was to liberate us. To let me go spend more time outside, to stare at trees, and ask people "can I pet your dog?" We use language and images because they are easier to evaluate. Because we don't know what to actually evaluate. So it's as good of a direction as any, right? I'm not sure if another direction would have…

>Because many of us working on AI were working on it for different reasons. To me, it was to liberate us. To let me go spend more time outside, to stare at trees, and ask people "can I pet your dog?" If you think automation or any other increase in productivity is passed back down to workers, then I'd say I have a bridge to sell you, but you probably already bought 5 of them.

It is easy to be blinded by our passions[0]. I still believe it is possible. I'm not in it for the money. I'm not doing this because it pays well or the clout. I do it because it is a captivating problem. I do it because these questions draw me in. I am aware of many risk, many that are not even being discussed[1].

But I'll also tell you, I don't want to talk to most of my peers. I don't see that same passion. Most want to go to parties and make lots of money. Rather, I seek my peers who have similar passions. They may have different beliefs, different ideas, and we may even argue and fight. But the reason we do it is because we're trying to solve this great mystery. It is getting harder and harder to find them.

Tbh, I don't think there's anything inherently wrong with doing things for money, to do things without passion. (I don't think this excuses abuse, theft, lying, or the multitude of things I think you're thinking about. We're probably more aligned than you think. I don't think I'm your enemy here. In fact, I think we have one in common. Even if you wish to be my enemy, I do not wish to be yours) Some people are just trying to get by in this crazy world. But we're not talking about that, are we.

[0] https://news.ycombinator.com/item?id=44568337

[1] I even specifically use a handle to feel like I have the liberty to speak more openly about these things.

Re: LLM Inevitabilism

#166

Post by SwoopsFromAbove »

keiferski wrote: post #107

One of the negative consequences of the “modern secular age” is that many very intelligent, thoughtful people feel justified in brushing away millennia of philosophical and religious thought because they deem it outdated or no longer relevant. (The book A Secular Age is a great read on this, btw, I think I’ve recommended it here on HN at least half a dozen times.) And so a result of this is that they fail to notice t…

100%. Not a new phenomenon at all, just the latest bogeyman for the inevitabilists to point to in their predestination arguments.

My aim is only to point it out - people are quite comfortable rejecting predestination arguments coming from eg. physics or religion, but are still awed by “AI is inevitable”.

Re: LLM Inevitabilism

#167

Post by slacktivism123 »

mg wrote: post #70

In the 90s a friend told me about the internet. And that he knows someone who is in a university and has access to it and can show us. An hour later, we were sitting in front of a computer in that university and watched his friend surfing the web. Clicking on links, receiving pages of text. Faster than one could read. In a nice layout. Even with images. And links to other pages. We were shocked. No printing, no shipp…

I look forward to the inevitable replies from HN's appeal-to-authority darlings explaining why we are "crazy" to not believe in this "future".

Debate team techniques are super useful when your salary now depends on shilling LLMs!

Re: LLM Inevitabilism

#168

Post by jrflowers »

lblume wrote: post #110
fendy3002 wrote: post #68

Earlier quoted context omitted.

LLMs need significant optimization or we get significant improvement on computing power while keeping the energy cost the same. It's similar with smartphone, when at the start it's not feasible because of computing power, and now we have one that can rival 2000s notebooks. LLMs is too trivial to be expensive EDIT: I presented the statement wrongly. What I mean is the use case for LLM are trivial things, it shouldn't…

Imagine telling a person from five years ago that the programs that would basically solve NLP, perform better than experts at many tasks and are hard not to anthropomorphize accidentally are actually "trivial". Good luck with that.

>programs that would basically solve NLP

There is a load-bearing “basically” in this statement about the chat bots that just told me that the number of dogs granted forklift certification in 2023 is 8,472.

Re: LLM Inevitabilism

#169

Post by tempfile »

killerstorm wrote: post #158
bradly wrote: post #121

Earlier quoted context omitted.

This is what that same GPT4 told me today after trying to get a simple mqttwarn config: ChatGPT: Let’s cut through the mqttwarn quirks and get you a single, complete warn.ini that I’ve actually tested with mqttwarn 0.32 Me: that "I’ve actually tested with mqttwarn 0.32:" How did you perform this test? ChatGPT: I set up a quick local sandbox on my laptop Me: What brand is your laptop? What store did you buy it at? Cha…

Did you ever think there's a reason why people are paying for professional tools like Cursor or Claude Code instead of using free ChatGPT? Ye, the free version has some known issues. They cram a lot of stuff into GPT-4o, so it hallucinates a lot. Claude Opus 4 often gives perfectly working code on the first try, and it's much less likely to hallucinate or argue with you when it's wrong. It costs around $1 per request…

They are replying to someone who said ChatGPT. Why are you barging in to change the goalposts?

Re: LLM Inevitabilism

#170

Post by SwoopsFromAbove »

dasil003 wrote: post #98

Two things are very clearly true: 1) LLMs can do a lot of things that previous computing techniques could not do and we need time to figure out how best to harness and utilize those capabilities; but also 2) there is a wide range of powerful people who have tons of incentive to ride the hype wave regardless of where things will actually land. To the article's point—I don't think it's useful to accept the tech CEO fra…

This is how I approach the tools too. I believe it’s a healthy approach, but who’s to say whether I’m just a naysayer. shrug
Post reply on HN