It's hard to give this take serious thought when just today I built a bunch of utility software I'd been putting off due to lack of time and domain specific knowledge, all by working together with an LLM. Takes some time and back and forth, but it works and does the things I need. It doesn't need to get everything right nor to get it right the first time, humans don't tick either of those boxes either. The value add is already clearly there and it's most likely going to improve over time.
You could make the same quip about nearly any piece of software, yet most everyone on HN still has a business or a job. If it exists, but I can't find it or it's too cumbersome to set up or use, it's functionally as good to me as it not existing.
Another way to look at this amazing performance improvement for you in making small tools is you’re just using bits of other people’s work that were lying around on the internet, without licensing, permission, or attribution.
It’s different because licensed code on github and other websites is different from freely given code snippets on stack overflow.
A blog post is not permission to reproduce code with small changes. Neither is posting it under an open source license which usually require attribution.
That’s funny, I guess everyone just ignores that, but legally it means AI should not be trained on it either since it requires attribution, I wonder was it always that way?
I think AI companies are just trying to ignore IP laws entirely, perhaps they’ll get away with it but there is bound to be a case where a small tool or game is 90% identical to a source reference at some point given the way these LLMs work.
> this is just what it predicted to be most fitting from it's training data
and your own prediction, from your own brain, is doing something quite similar i would imagine.
The mechanism behind how it works is irrelevant. The results can be judged on its own. People used to judge chess engines as tho they are just merely searching and trying every possibility, and it's not "really playing chess", like humans' intuitions. And yet, by just doing searching and almost brute force, it produces a better chess game than any humans can.
And LLM are still in the early days. It's only been around for less than 3 years.
You’re getting flamed (haven’t used that word in a while) a bit here but have a point. AI is useful but not the civilization disrupting tech we were promised or warned. To me it’s like conversational stack overflow.
I am slightly astonished someone makes these sorts of comments in 2025. AI has been remarkably useful for many many things across many industries; I’m curious what you think
> AI has been remarkably useful for many many things across many industries
This is really a statement of faith more than a statement of fact. I think you and many others believe this to be true without much concrete evidence. For work, I help companies adopt AI driven solutions. Sometimes it makes things a little better, sometimes it makes things worse. I've yet to see a project use LLMs in the transformative way that many AI optimists put forward. Don't get me wrong, I find tools like Claude and ChatGPT to be fascinating and useful for looking up all kinds of information. I can't really say if we're just scratching the surface or if we've dug ourselves into rut with the present state of LLMs. The firsthand evidence I've seen and verified points more to the latter, but things are changing fast in this area. I'm excited to see what's around the corner.
Simple google searches did this too, it still takes an true intelligence to apply it.
At best its a ridiculously inefficient search engine that sits atop the million corpses of failed models
Not OP, but I use AI tools and sometimes it’s great, sometimes they’ll distractingly lead you in circles, and other times they completely shit the bed.
Luckily, I’m using AI tools to do things that I am capable of doing without AI so I can tell which path I’m going down pretty quickly.
So, letting experts augment their skills with AI is something that can work depending on the specific task. The nice thing is that an expert is able to see errors pretty quickly and determine that this task is a mismatch for the AI.
The problem is that AI is being sold as a magic solution so people never need to develop expertise in the first place. Blind trust in a system that is often confidently incorrect will lead to problems
Dunno, on one hand, I think LLMs are somewhat of a dead end. We trained them on all human knowledge, and they're still not great. Useful enough, but a tad underwhelming versus the hype.
On the other hand, with all the money flooded into AI, all the hardware that's been produced and bought, it's reignited the entire AI industry (that mostly died in the 80's) and there is a chance for innovation beyond LLMs.
LLMs are a calculator for language, but not for reasoning.
When you realize that language and reasoning are in fact two separate skills, only one of which an LLMs is good at, they make much more sense.
Until now, language skill and reasoning skill have been correlated - people with greater skill using language are usually better skilled at reasoning. Put another way, we typically discount poorly written material regardless of it's actual content.
LLMs turn this on its head - great at language, poor at reasoning. So the crutch, the heuristic, we used before no longer applies. We MUST recognize that language ability and reasoning ability are now independent.
