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

The last six months in LLMs in five minutes

simonwillison.net

Post reply on HN
Print view View original

621–630 of 631 posts

Re: The last six months in LLMs in five minutes

#621

Post by eudamoniac »

ajam1507 wrote: post #595
eudamoniac wrote: post #558

Earlier quoted context omitted.

I'm convinced that the polarization is that one's impression of AI has a direct 1:1 mapping with one's previous level of skill and sensitivity to quality. Most people are by definition average and they are impressed. Is there anyone in the industry noted for their skill, quality, and taste, e.g. Jonathon Blow, who is impressed and thinks the AI is really good? I haven't seen any. In my personal circle, the best devs…

This is contradicted by the amount of AI use at top tech firms.

No, it would be contradicted by the sentiment among devs at top tech firms, but I don't know what the sentiment is. I do know they are being forced to use AI at peril of termination such that their use of AI is a non signal.

Re: The last six months in LLMs in five minutes

#622

Post by Miraste »

hollowturtle wrote: post #555
Miraste wrote: post #533

Earlier quoted context omitted.

What novel ideas are you thinking of? In my experience there are very few games with novel software engineering. New gameplay mechanics or story or art or design, sure, but they're generally built with very old and standardized patterns.

If what you say it's true, where are the great vibe coded games? Or great non vibe coded games, the killer apps that sold millions on steam?

There's no way to prove it one way or the other. I would be willing to bet money any recent AAA game was made with AI code in it somewhere, though.

Re: The last six months in LLMs in five minutes

#623

Post by okamiueru »

JohnKemeny wrote: post #461
vessenes wrote: post #291

Earlier quoted context omitted.

Spicy — why does it show ultimate non-understanding?

because success comes from reproducing a memorized pattern rather than transferable reasoning? At the same time failure proves little because most humans also could not manually create a correct SVG of a pelican riding a bicycle. What is it exactly that such a test is testing? In which situation would you measure the "competence" of a human being by asking them to write an SVG of a pelican riding a bicycle?

> most humans also could not manually create a correct SVG of a pelican riding a bicycle.

Most humans absolutely can write this with a suitable vector graphics tool such as inkscape or illustrator.

Surely, you're not suggesting that a fair comparison would be using a text editor?

If so, would you suggest an equivalent raster based task would only be fair, if the human would manually assigning RGB values to each pixel?

Re: The last six months in LLMs in five minutes

#624

Post by alexwwang »

BOOSTERHIDROGEN wrote: post #112
alexwwang wrote: post #109

Earlier quoted context omitted.

As a former data scientist, I started to use code agent 3 monthes ago. Before that, I use chat completion on web. Now, I nearly do everything which outputs documents with code agent.

Can you give a sanitized example or a hypothetical scenario of what you mean by “output documents with code agents”? Thanks.

All the documents that were typed with a keyboard before, now can be created by code agents with properly designed and implemented prompts and skills.

I generate my blog with this method and you can refer to: https://blog.chuanxilu.net/en/

I am responsible for all the contents but the process of those essays and reports are first generated by prompts that embody my ideas, thoughts and facts I encountered.

Re: The last six months in LLMs in five minutes

#625

Post by alexwwang »

schnitzelstoat wrote: post #149
BOOSTERHIDROGEN wrote: post #112

Earlier quoted context omitted.

Can you give a sanitized example or a hypothetical scenario of what you mean by “output documents with code agents”? Thanks.

I’m not him, but I’ve started using them to do the analysis (SQL, Python etc.) and then output the report as Quarto HTML which can be hosted on GitHub Pages. It works well for this analysis style work. Once I was going to send some figures to leadership so I checked the queries myself and not only had it done it correctly, but it had also included a lot of sanity checks with other places in the database which as a hu…

Yes dude. You understand fully what I mean.

Re: The last six months in LLMs in five minutes

#626

Post by jackzhuo »

kvakkefly wrote: post #30
Insanity wrote: post #10

I wonder how much the 'inflection point' is a thing vs marketing. I'm sure the models got somewhat better, but even now when I'm trying to 'vibe code' a game with the latest models (combination of Codex w/ gpt5.5 and gpt5.3-codex), they really do struggle. They definitely get something barebones up and running, but it's far from a fully fledged application.

I remember this very clearly myself. Before opus 4.5, I was doing a lot of hand holding and was coding a lot myself, but I have not written code since that day more or less. I did write some stuff myself just to learn how the enigma encryption machine worked, so wrote myself to learn. But professionally, I stopped coding in November.

Same experience here. I now think AI writes much better code than me. So I shifted my focus to finding requirements, analyzing possibilities, and making good plans.

Re: The last six months in LLMs in five minutes

#627

Post by krzyk »

bloppe wrote: post #178
krzyk wrote: post #88

Earlier quoted context omitted.

It is sad. I like programming, if I couldn't do it and had to write text (which I do hate, I'm not a writer) it would be make quite a sad world.

A pattern I've settled into is to write code but leave a TODO for every narrow thing I want the LLM to do for me. Then just tell the agent to fix the todos. It's often faster and easier to give "instructions" this way

I never thought about that, good idea.

Re: The last six months in LLMs in five minutes

#628

Post by Glohrischi »

forlorn_mammoth wrote: post #464
Glohrischi wrote: post #402

Earlier quoted context omitted.

I had a really fun day yesterday because anthropics limits on their normal 20$ subscription allowed me to play around for the whole day without hitting a limit. Its 'production' code because its a small browser game which has very small to 0 requirements on security and being perfect but high requirements on 'ever even doing this' and 'fun'. The code it generated hat 0 compiletime errors. I was able to descripe 10 th…

> The code it generated hat 0 compiletime errors And no spelling errors either! Also, > Really? What duplication did you actually find? I count a few small ones in buildMounts and ReadPrompt, maybe 20 lines or so, but hardly anything worthy of such an epithet >> embedding-shape 1 hour ago | root | parent | next [–] >>The duplication I'm seeing isn't just "same text repeated" but structural duplication. Doing a quick…

I'm not a native english speaker and when i mentioned that i might use LLM for fixing spellings, people argued about the use of LLM. So spelling error yes/no?

I do not understand the quote you rference at all tbh?

Re: The last six months in LLMs in five minutes

#629

Post by gib444 »

x86cherry wrote: post #126
gib444 wrote: post #89

Starting from zero today, how would someone quickly get upto speed with the latest and greatest AI tooling on an extremely limited budget? Is the only choice to pay for the "max" plans? Or just read so much about it that you bs your way through an interview and then use the company's resources? Simon, I'm curious too how much you invest each month researching all the latest and great AI tech?

Opencode has free access to Qwen 3.6 and Deepseek v4 Flash right now. They're on par with Claude and Codex imo - when you still design architecture and know what the output should be. Claude and GPT 5.5 need less guidance with vibe coding, but we're not yet at a point where that's sustainable anyway even with those models.

Thanks, playing with Opencode now. It just wrote a half-decent Android small app. Pretty good so far!

Re: The last six months in LLMs in five minutes

#630

Post by bluegatty »

nl wrote: post #613
bluegatty wrote: post #609

Earlier quoted context omitted.

Sure, something like that. Note like the examples you posted.

I've very confused. The SVGs show the beak, wing, tail, feet and body as though viewed from directly underneath. They look similar to the photo, but meet the instructions better ("from underneath"). What are you expecting exactly?

They don't look anything like the photo.

They're not 'oblique' - they're 'squared' views and none of the anatomy looks appropriately adjusted.

The model has no ability to 'rotate a figure in 3d space' and conceptualize how all of the elements work together.

It's 'pattern matching'.

This is the 'great intuition' for how LLMs work - it's not perfect because a lot of 'synthetic reasoning' can be done obviously.

And they probably never will, LLMs are not the right thing for this kind of task.

Think about how they can investigate massive code-bases and find arcane bugs - but cant draw a duck from arbitrary oblique angles etc.

That said, with enough examples they probably could.

Post reply on HN