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

Andrej Karpathy's talk on the future of the industry

donnamagi.com

Post reply on HN
Print view View original

51–60 of 193 posts

Re: Andrej Karpathy's talk on the future of the industry

#51

Post by Frummy »

kaladin-jasnah wrote: post #43

Tangentially related, but it boggles my mind this guy was badmephisto, who made a quite famous cubing tutorial website that I spent plenty of time on in my childhood.

Totally not a supervillain

"Q: What does your name (badmephisto) mean?

A: I've had this name for a really long time. I used to be a big fan of Diablo2, so when I had to create my email address username on hotmail, i decided to use Mephisto as my username. But of course Mephisto was already taken, so I tried Mephisto1, Mephisto2, all the way up to about 9, and all was taken. So then I thought... "hmmm, what kind of chracteristic does Mephisto posess?" Now keep in mind that this was about 10 years ago, and my English language dictionary composed of about 20 words. One of them was the word 'bad'. Since Mephisto (the brother of Diablo) was certainly pretty bad, I punched in badmephisto and that worked. Had I known more words it probably would have ended up being evilmephisto or something :p"

Re: Andrej Karpathy's talk on the future of the industry

#52

Post by Aeroi »

TL;DR: Karpathy says we’re in Software 3.0: big language models act like programmable building blocks where natural language is the new code. Don’t jump straight to fully autonomous “agents”—ship human-in-the-loop tools with an “autonomy slider,” tight generate-→verify loops, and clear GUIs. Cloud LLMs still win on cost, but on-device is coming. To future-proof, expose clean APIs and docs so these models (and coming agents) can safely read, write, and act inside your product.

Re: Andrej Karpathy's talk on the future of the industry

#53

Post by jasonjmcghee »

steveBK123 wrote: post #46
jolan wrote: post #39

Earlier quoted context omitted.

I think he meant the audio output itself sounds like an infomercial. https://www.youtube.com/watch?v=gfr4BP4V1R8

Correct, yes. That NotebookLM podcast was like the most unpleasant way I can imagine to consume content. Reading transcripts of live talks is already pretty annoying because it's less concise than the written word. Having it re-expanded by robot-voice back to audio to be read to me just makes it even more unpleasant. Also sort of perverse we are going audio->transcript->fake audio. "YC has said the official video wil…

I have the same perspective - to such a degree that any time I see someone post a notebooklm I wonder if it's paid advertising. Every time I've tried it on something like a whitepaper etc. it just makes stuff up or says things that are kind of worthless. Reminds me of ChatGPT 3.5 in terms of quality of the presented content.

The voices sounded REALLY good the first time I used it. But then sounded exactly the same every time after that and became underwhelmed.

Re: Andrej Karpathy's talk on the future of the industry

#54

Post by arkj »

>Software 2.0 are the weights which program neural networks. >I think it's a fundamental change, is that neural networks became programmable with large libraries... And in my mind, it's worth giving it the designation of a Software 3.0.

I think it's a bit early to change your mind here. We love your 2.0, let's wait for some more time till th e dust settles so we can see clearly and up the revision number.

In fact I'm a bit confused about the number AK has in mind. Anyone else knows how he arrived at software 2.0?

I remember a talk by professor Sussman where he suggest we don't know how to compute, yet[1].

I was thinking he meant this,

Software 0.1 - Machine Code/Assembly Code Software 1.0 - HLLs with Compilers/Interpreters/Libraries Software 2.0 - Language comprehension with LLMs

If we are calling weights 2.0 and NN with libraries as 3.0, then shouldn't we account for functional and oo programming in the numbering scheme?

[1] https://www.youtube.com/watch?v=HB5TrK7A4pI

Re: Andrej Karpathy's talk on the future of the industry

#56

Post by steveklabnik »

afiodorov wrote: post #37
andai wrote: post #36

Earlier quoted context omitted.

The last 10% always takes 1000% of the time...

I am not saying rewrites are always warranted, but I think LLMs change the cost-benefit balance considerably.

I am with you on this. I'm very much not sure about rewrites, but LLMs do change the cost-benefit balance of refactorings considerably, for me. Both in a "they let me make a more informed decision about proceeding with the refactoring" and "they are faster at doing it than I am".

Re: Andrej Karpathy's talk on the future of the industry

#57

Post by serial_dev »

alganet wrote: post #42

> imagine changing it and programming the computer's life > imagine that the inputs for the car are on the bottom, and they're going through the software stack to produce the steering and acceleration > imagine inspecting them, and it's got an autonomy slider > imagine works as like this binary array of a different situation, of like what works and doesn't work -- Software 3.0 is imaginary. All in your head. I'm kidd…

I tried to imagine all that he described and felt literally nothing. If he wants to hype AI, he should find his Steve Jobs.

Re: Andrej Karpathy's talk on the future of the industry

#58

Post by tayo42 »

no_wizard wrote: post #25
umeshunni wrote: post #18

Earlier quoted context omitted.

> I do think that open source models will eventually catch up to closed source ones It felt like that was the direction for a while, but in the last year or so, the gap seems to have widened. I'm curious whether this is my perception or validated by some metric.

This was how early browsers felt too, the open source browser engines were slower at adapting than the ones developed by Netscape and Microsoft, but eventually it all reversed and open source excelled past the closed source software. Another way to put it, is that over time you see this, it usually takes a little while for open source projects to catch up, but once they do they gain traction quite quickly over the cl…

Those were way simpler projects in the beginning when that happened. Like do you think a new browser would catch up today chrome now?

Re: Andrej Karpathy's talk on the future of the industry

#60

Post by msgodel »

serial_dev wrote: post #57
alganet wrote: post #42

> imagine changing it and programming the computer's life > imagine that the inputs for the car are on the bottom, and they're going through the software stack to produce the steering and acceleration > imagine inspecting them, and it's got an autonomy slider > imagine works as like this binary array of a different situation, of like what works and doesn't work -- Software 3.0 is imaginary. All in your head. I'm kidd…

I tried to imagine all that he described and felt literally nothing. If he wants to hype AI, he should find his Steve Jobs.

It was easy for me to see and it's incredible. Maybe I should be launching a startup.
Post reply on HN