Hacker Newsnew | past | comments | ask | show | jobs | submit | kaeluka's commentslogin

Maps grounding is an amazing interface for Google Maps - „I’m at x, need to be at y by 5pm. Find me a route that’s walkable and has a good vegan restaurant and a toy store“

I’ve been toying with using an llm to compile to smtlib and solving with z3. It works quite well, although I’m not quite sure how practical (large programs, reliably good runtime) I can make it.

I thought the same then. But the funny thing is that today, it has become a lot easier to recognize the frontier models as not human. All the load bearing and not x but y, etc… weird


This is a tell of LLMs but it's not universal. I use ChatGPT extensively and I don't often get obvious nonsense any more.

I'd figure out that it's an LLM because it's effectively superhuman. Taking that away I'm not so sure I'd be able to tell


The ultimate tell is still the sycophancy. You agree with the suggestion output by the LLM but if it detects even a slight pushback it will completely reverse the previous suggestion. Only way to make it more obvious would be to have the LLM grovel and beg.

If these things have consciousness then we are committing sadism on a massive scale.


It feels like it's getting better at that too. It will push back against obvious nonsense a lot stronger. But you are right that it holds opinions quite a bit less strenuously than humans.


It's a smart model, but it has huge troubles staying on task... it constantly builds things I didn't ask it to.

Might have its uses in a tightly controlled harness, but for coding I'm switching back to a mix of K3/GLM-5.2/Deepseek V4 Flash 0731


i came back to say: i've given this another chance and while my point still stands: it's actually making up for that by being a good thinking partner.

the latest project i've been working on this has been really good. So I may have judged too quickly.


Interesting!

GLM 5.2 is slower for sure (although they offer a fast version), and it's more expensive. But in my experience, it's universally better than Deepseek V4-flash-0731. Don't get me wrong, the new Flash version is amazing.

But the use cases I have looked at are about source code understanding, bug finding, etc. - GLM 5.2 is clearly better.

I think by using some prompt engineering, you will probably be able to close this gap, but some extra work is needed.

And I'll say it again: the new Flash version is amazing. I love it. That level of intelligence for the price is unprecedented, and the fact that it's open weights and runs locally makes me genuinely happy.


great, i'll try this. something like this has on my list and i'm super curious :)


They're trying to make the product sticky while they're still subsidizing the subscription price. The plan is to raise prices when you're addicted.


it's sort of good at thinking, writing specs, etc.. Also debugging. But as a coder: I see no advantage to opus 4.6 and I preferred sonnet most times already over opus 4.6.


Thanks for the tip, started out really interesting!


Even if you don't care about fusion itself, it is highly interesting in that it covers some of the engineering that went into building that thing.


And the guests are perfect. Competent and entertaining at the same time.


That blog entry is well worth reading. It put a smile on my face! :-)


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: