> But models are useful just like... all the regexes I never learned how to write and now never will!
Wait, does this mean I'm better at something than geohot? All that time spent learning regexps wasn't a waste!
I love LLMs, I hate hype
51–60 of 340 posts
Re: I love LLMs, I hate hype
#52I love LLMs too, but I am concerned about their cost. They are all still very subsidised. Is there any guarantee that I'll be able to run a Opus 4.8-level model on my personal computer before the big AI labs decide to hike up the prices?
> They are all still very subsidised. I think the opposite: I think the frontier labs have good margins on their inference unit costs. We can already see what it costs to run near frontier-size models. There are independent business pivoting to serving these models at reasonable prices and they're competing on OpenRouter for costs much lower than frontier labs. > Is there any guarantee that I'll be able to run a Opus…
Interestingly enough, geohot also has an article covering this: https://geohot.github.io//blog/jekyll/update/2026/06/18/pric...
Re: I love LLMs, I hate hype
#53Your SF hate isn't a good look. There are many things to be critical about but shoehorning an entire metro into the echo-chamber you're supposedly beyond yet can't help but orient your entire world view as the anti-SF-tech-bro all while running a startup and discussing AI on HN. TLDR: SF is more than Paul Graham worship parties. EDIT: Think I'm being misunderstood! author goes out of his way to blame shitty San Franc…
> shoehorning an entire metro into the echo-chamber you're supposedly beyond
The SF metro is possibly the worst in the entire world in terms of CoL vs QoL.
It has a higher proportion of unsheltered population living on the streets than almost any city outside of Africa except Manila and possibly Dhaka
Re: I love LLMs, I hate hype
#54As soon as we started unironically calling LLMs "AI" we went down the hype path. That has plenty of downsides, like stressing out the entire world and attracting cryptocurrency bros, but also the major upside massive of funding/acceleration. So far, all we have is more software running on computers. It's powerful, and it's amazing, but it's not magic. Calling it "AI" was possibly a net-negative but we don't know yet.
"It's powerful, and it's amazing, but it's not magic" But since its creators and as of my knowledge everyone else totally did not see it coming, that you can now give a vague prompt full of spelling errors - and get returned a working program - I would say it is pretty close to magic (as in we don't really understand why it works so good). I also don't see how you cannot call it AI. Especially since simple chess engi…
Yes, LLM agents are "magic" in the sense that
"any sufficiently advanced technology is indistinguishable from magic"[1]
But it's not actually magic. Technical people understand that it's just software running on computers.
Re: I love LLMs, I hate hype
#55[flagged]
Re: I love LLMs, I hate hype
#56Earlier quoted context omitted.
> They are all still very subsidised. I think the opposite: I think the frontier labs have good margins on their inference unit costs. We can already see what it costs to run near frontier-size models. There are independent business pivoting to serving these models at reasonable prices and they're competing on OpenRouter for costs much lower than frontier labs. > Is there any guarantee that I'll be able to run a Opus…
Interestingly enough, geohot also has an article covering this: https://geohot.github.io//blog/jekyll/update/2026/06/18/pric...
That's commentary on company valuations.
Token prices are going down. Competition is global. A company could choose to keep their API prices high, but if another company comes in at 1/10th the price for 95% of the performance then they won't have many customers.
Re: I love LLMs, I hate hype
#57I felt the same way in 2024-2025. Then Sonnet 4 was released, and things started feeling different. Opus 4.5 was another step change for me. Everything feels like it's accelerating, and timelines are getting crunched. I guess in some ways I envy OP, who would "bet everything" against ASI - the truth is I don't know, and I don't think anyone knows, where this ends.
He didn't say he bets everything against ASI, he said he bets everything against ASI being a flash of light in the sky which destroys our chance of getting access to the wealth it creates.
Re: I love LLMs, I hate hype
#58I get it, I want to agree, I really do like the “this is a new tool in the toolkit of the professional software craftsperson” argument…
…but consider: the Q-tip. “Don’t use it to clean your ears”, but for most people that’s all they want to do with it, and empirical observation indicates that this dynamic results in either “using Q-tips irresponsibly” or “not using Q-tips”, with “uses Q-tips properly” being a small-to-vanishing proportion of the whole.
Re: I love LLMs, I hate hype
#59Earlier quoted context omitted.
Interestingly enough, geohot also has an article covering this: https://geohot.github.io//blog/jekyll/update/2026/06/18/pric...
That's commentary on company valuations. Token prices are going down. Competition is global. A company could choose to keep their API prices high, but if another company comes in at 1/10th the price for 95% of the performance then they won't have many customers.
You’re right, my bad, I read that too quickly
Re: I love LLMs, I hate hype
#60This guy is sooooo annoying with his stale takes. This is what he wrote before. > I’m calling it now, the adoption of AI agents into software development will be one of the most costly mistakes in the field’s history. Agents cannot program, and it’s taking longer and longer to realize that they can’t. Now he's writng > I love the progress. I’m so excited for the new LLMs, self driving cars, video generation models, a…
Both can be true and I have both opinions also in me. Love the progress, worry about the consequences of not being careful with it.
He does say in this post:
> I’m getting better at using them and get some boost from the models. It is a new skill, and it’s not like I haven’t constantly been trying them. You have to be really careful, they can increase cognitive fatigue, and all the vibe coded stuff is still slop (where’s all this new magical software that the productivity improvements should imply?).