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

Andrej Karpathy: Software in the era of AI [video]

youtube.com

Post reply on HN
Print view View original

591–600 of 827 posts

Re: Andrej Karpathy: Software in the era of AI [video]

#591

Post by skydhash »

variadix wrote: post #429
mistersquid wrote: post #357

Earlier quoted context omitted.

> I can recognize images in one look. > How about that 400 Line change that touches 7 files? Karpathy discusses this discrepancy. In his estimation LLMs currently do not have a UI comparable to 1970s CLI. Today, LLMs output text and text does not leverage the human brain’s ability to ingest visually coded information, literally, at a glance. Karpathy surmises UIs for LLMs are coming and I suspect he’s correct.

The thing required isn’t a GUI for LLMs, it’s a visual model of code that captures all the behavior and is a useful representation to a human. People have floated this idea before LLMs, but as far as I know there isn’t any real progress, probably because it isn’t feasible. There’s so much intricacy and detail in software (and getting it even slightly wrong can be catastrophic), any representation that can capture sai…

There’s no visual model for code as code isn’t 2d. There’s 2 mechanism in the turing machine model: a state machine and a linear representation of code and data. The 2d representation of state machine has no significance and the linear aspect of code and data is hiding more dimensions. We invented more abstractions, but nothing that map to a visual representation.

Re: Andrej Karpathy: Software in the era of AI [video]

#592

Post by digianarchist »

belter wrote: post #91

Painful to watch. The new tech generation deserves better than hyped presentations from tech evangelists. This reminds me of the Three Amigos and Grady Booch evangelizing the future of software while ignoring the terrible output from Rational Software and the Unified Process. At least we got acknowledgment that self-driving remains unsolved: https://youtu.be/LCEmiRjPEtQ?t=1622 And Waymo still requires extensive human…

In his defense he clearly articulated that meaningful change has not yet been achieved and could be a decade away. Even pointing to specific examples of LLMs failing to count letters and do basic arithmetic.

What I find absent is where do we go from LLMs? More hardware, more training. "This isn't the scientific breakthrough you're looking for".

Re: Andrej Karpathy: Software in the era of AI [video]

#593

Post by magicloop »

I think this is a brilliant talk and truly captures the "zeitgeist" of our times. He sees the emergent patterns arising as software creation is changing.

I am writing a hobby app at the moment and I am thinking about its architecture in a new way now. I am making all my model structures comprehensible so that LLMs can see the inside semantics of my app. I merely provide a human friendly GUI over the top to avoid the linear wall-of-text problem you get when you want to do something complex via a chat interface.

We need to meet LLMs in the middle ground to leverage the best of our contributions - traditional code, partially autonomous AI, and crafted UI/UX.

Part of, but not all of, programming is "prompting well". It goes along with understanding the imperative aspects, developing a nose for code smells, and the judgement for good UI/UX.

I find our current times both scary and exciting.

Re: Andrej Karpathy: Software in the era of AI [video]

#594

Post by skydhash »

Workaccount2 wrote: post #532
whilenot-dev wrote: post #514

I watched Karpathy's Intro to Large Language Models [0] not so long ago and must say that I'm a bit confused by this presentation, and it's a bit unclear to me what it adds. 1,5 years ago he saw all the tool uses in agent systems as the future of LLMs, which seemed reasonable to me. There was (and maybe still is) potential for a lot of business cases to be explored, but every system is defined by its boundaries nonet…

The fundamental mistake I see is people applying LLMs to the current paradigm of software; enormous hulking codebases made to have as many features as possible to appeal to as many users as possible. LLMs are excellent at helping non-programmers write narrow use case, bespoke programs. LLMs don't need to be able to one-shot excel.exe or Plantio.apk so that Christine can easily track when she watered and fed her plant…

The thing is that there’s a need to integrate all these little tools because the problems they solve is part of the same domain. And that’s where problems lie. Something like Excel have an advantage as being a common platform for both data and procedures. Unix adopted text and pipes for integration.

Re: Andrej Karpathy: Software in the era of AI [video]

#595

Post by simplify »

otabdeveloper4 wrote: post #545
kordlessagain wrote: post #414

Earlier quoted context omitted.

This thread perfectly captures what Karpathy was getting at. We're witnessing a fundamental shift where the interface to computing is changing from formal syntax to natural language. But you can see people struggling to let go of the formal foundations they've built their careers on.

> We're witnessing a fundamental shift where the interface to computing is changing from formal syntax to natural language. People have said this every year since the 1950's. No, it is not happening. LLMs won't help. Writing code is easy, it's understanding the problem domain is hard. LLMs won't help you understand the problem domain in a formal manner. (In fact they might make it even more difficult.)

Let's be real, people have said similar things about AI too. It was all fluff, until it wasn't.

Re: Andrej Karpathy: Software in the era of AI [video]

#596

Post by throwaway314155 »

redbell wrote: post #581
practal wrote: post #50

Great talk, thanks for putting it online so quickly. I liked the idea of making the generation / verification loop go brrr, and one way to do this is to make verification not just a human task, but a machine task, where possible. Yes, I am talking about formal verification, of course! That also goes nicely together with "keeping the AI on a tight leash". It seems to clash though with "English is the new programming l…

> "English is the new programming language." For those who missed it, here's the viral tweet by Karpathy himself: https://x.com/karpathy/status/1617979122625712128

Referenced in the video of course. Not that everyone should watch a 40 minute long video before commenting but his reaction to the "meme" that vibe coding became when his tweet was intended as more of a shower thought is worth checking out.

Re: Andrej Karpathy: Software in the era of AI [video]

#597

Post by throwaway314155 »

karpathy wrote: post #57
hgl wrote: post #51

It’s fascinating to think about what true GUI for LLM could be like. It immediately makes me think a LLM that can generate a customized GUI for the topic at hand where you can interact with in a non-linear way.

Fun demo of an early idea was posted by Oriol just yesterday :) https://x.com/OriolVinyalsML/status/1935005985070084197

I would bet good money that many of the functions they chose not to drill down into (such as settings -> volume) do nothing at all or cause an error.

It's a fronted generator. It's fast. That's cool. But is being pitched as a functioning OS generator and I can't help but think it isn't given the failure rates for those sorts of tasks. Further, the success rates for HTML generation probably _are_ good enough for a Holmes-esque (perhaps too harsh) rugpull (again, too harsh) demo.

A cool glimpse into what the future might look like in any case.

Re: Andrej Karpathy: Software in the era of AI [video]

#598

Post by polivier »

practal wrote: post #432
gylterud wrote: post #340

Earlier quoted context omitted.

Why? Why would the language used to express proof of correctness have anything to do with English? English was not developed to facilitate exact and formal reasoning. In natural language ambiguity is a feature, in formal languages it is unwanted. Just look at maths. The reasons for all the symbols is not only brevity but also precision. (I dont think the symbolism of mathematics is something to strive for though, we…

My answer is already in my previous comment: if you have two formal languages to choose from, you want the one closer to natural language, because it will be easier to see if informal and formal statements match. Once you are in formal land, you can do transformations to other formal systems as you like, as these can be machine-verified. Does that make sense?

> if you have two formal languages to choose from, you want the one closer to natural language

Given the choice I'd rather use Python than COBOL even though COBOL is closer to English than Python.

Re: Andrej Karpathy: Software in the era of AI [video]

#599

Post by hellovai »

solaire_oa wrote: post #411
miki123211 wrote: post #38

Earlier quoted context omitted.

What do you think about structured outputs / JSON mode / constrained decoding / whatever you wish to call it? To me, it's a criminally underused tool. While "raw" LLMs are cool, they're annoying to use as anything but chatbots, as their output is unpredictable and basically impossible to parse programmatically. Structured outputs solve that problem neatly. In a way, they're "neural networks without the training". The…

I also think that structured outputs are criminally underused, but it isn't perfect... and per your example, it might not even be good, because I've done something similar. I was trying to make a decent cocktail recipe database, and scraped the text of cocktails from about 1400 webpages. Note that this was just the text of the cocktail recipe, and cocktail recipes are comparatively small. I sent the text to an LLM fo…

have you tried schema-aligned parsing yet?

the idea is that instead of using JSON.parse, we create a custom Type.parse for each type you define.

so if you want a:

   class Job { company: string[] }
And the LLM happens to output:

   { "company": "Amazon" }
We can upcast "Amazon" -> ["Amazon"] since you indicated that in your schema.

https://www.boundaryml.com/blog/schema-aligned-parsing

and since its only post processing, the technique will work on every model :)

for example, on BFCL benchmarks, we got SAP + GPT3.5 to beat out GPT4o ( https://www.boundaryml.com/blog/sota-function-calling )

Re: Andrej Karpathy: Software in the era of AI [video]

#600

Post by abdullin »

bayindirh wrote: post #355
abdullin wrote: post #303

Earlier quoted context omitted.

Exactly! This is why there has to be "write me a detailed implementation plan" step in between. Which files is it going to change, how, what are the gotchas, which tests will be affected or added etc. It is easier to review one document and point out missing bits, than chase the loose ends. Once the plan is done and good, it is usually a smooth path to the PR.

So you can create a more buggy code remixed from scraped bits from the internet which you don't understand, but somehow works rather than creating a higher quality, tighter code which takes the same amount of time to type? All the while offloading all the work to something else so your skills can atrophy at the same time? Sounds like progress to me.

Here is another way to look at the problem.

There is a team of 5 people that are passionate about their indigenous language and want to preserve it from disappearing. They are using AI+Coding tools to:

(1) Process and prepare a ton of various datasets for training custom text-to-speech, speech-to-text models and wake word models (because foundational models don't know this language), along with the pipelines and tooling for the contributors.

(2) design and develop an embedded device (running ESP32-S3) to act as a smart speaker running on the edge

(3) design and develop backend in golang to orchestrate hundreds of these speakers

(4) a whole bunch of Python agents (essentially glorified RAGs over folklore, stories)

(5) a set of websites for teachers to create course content and exercises, making them available to these edge devices

All that, just so that kids in a few hundred kindergartens and schools would be able to practice their own native language, listen to fairy tales, songs or ask questions.

This project was acknowledged by the UN (AI for Good programme). They are now extending their help to more disappearing languages.

None of that was possible before. This sounds like a good progress to me.

Edit: added newlines.

Post reply on HN