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

Andrej Karpathy – It will take a decade to work through the issues with agents

dwarkesh.com

Post reply on HN
Print view View original

101–110 of 1001 posts

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#101

Post by j45 »

mpalmer wrote: post #85
sosodev wrote: post #77

I think it's a shame that a 146 minute podcast released ~55 minutes ago has so much discussion. Everybody here is clearly just reacting to the title with their own biases. I know it's against the guidelines to discuss the state of a thread, but I really wish we could have thoughtful conversations about the content of links instead of title reactions.

Be fair; plenty of people transcribe and read podcasts, and/or summarize/excerpt them.

Summaries are great, but can be surface.

The brain processes and has insights differently experiencing it at conversation speed.

We might get what the conversation was that others had, but it can miss the mark for the listening and inner processing that leads to it's own gifts.

It's not about one or the other for me, usually both.

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#102

Post by zeroonetwothree »

somanyphotons wrote: post #76
Imnimo wrote: post #68

>What takes the long amount of time and the way to think about it is that it’s a march of nines. Every single nine is a constant amount of work. Every single nine is the same amount of work. When you get a demo and something works 90% of the time, that’s just the first nine. Then you need the second nine, a third nine, a fourth nine, a fifth nine. While I was at Tesla for five years or so, we went through maybe three…

This is an amazing quote that really applies to all software development

Well, maybe not all. I’ve definitely built CRUD UIs that were linear in effort. But certainly anything technically challenging or novel.

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#103

Post by throw54465665 »

flatline wrote: post #59
angiolillo wrote: post #43

"The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." - Edsger Dijkstra The debate about AGI is interesting from a philosophical perspective, but from a practical perspective AI doesn't need to get anywhere close to AGI to turn the world upside down.

I don’t even know what AGI is, and neither does anyone else as far as I can tell. In the parts of the video I watched, he cites several things missing which all have to do with autonomy: continual automated updates of internal state, fully autonomous agentic behavior, etc. I feel like GPT 3 was AGI, personally. It crossed some threshold that was both real and magical, and future improvements are relying on that basic…

Most humans do not even have a general intelligence! Many students are practically illiterate, and can not even read and understand book or manual!

We are approaching situation, where AI will make most decisions, and people will wear it as a skin suit, to fake competency!

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#105

Post by gowld »

zeroonetwothree wrote: post #97
Imnimo wrote: post #68

>What takes the long amount of time and the way to think about it is that it’s a march of nines. Every single nine is a constant amount of work. Every single nine is the same amount of work. When you get a demo and something works 90% of the time, that’s just the first nine. Then you need the second nine, a third nine, a fourth nine, a fifth nine. While I was at Tesla for five years or so, we went through maybe three…

When I worked at Facebook they had a slogan that captured this idea pretty well: “this journey is 1% finished”.

Copied from Amazon's "Day 1".

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#106

Post by IshKebab »

angiolillo wrote: post #43

"The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." - Edsger Dijkstra The debate about AGI is interesting from a philosophical perspective, but from a practical perspective AI doesn't need to get anywhere close to AGI to turn the world upside down.

Fortunately I haven't heard anyone make silly claims about stochastic parrots and the impossibility of conscious computers for quite a while.

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#108

Post by pixelpoet »

porphyra wrote: post #34
bcrosby95 wrote: post #31

Earlier quoted context omitted.

AI is an overloaded term. I took an AI class in 2001. We learned all sorts of algorithms classified as AI. Including various ML techniques. Under which included perceptrons.

back in the day alpha-beta search was AI hehe

As a young child in Indonesia we had an exceptionally fancy washing machine with all sorts of broken English superlatives on it, including "fuzzy logic artificial intelligence" and I used to watch it doing the turbo spin or whatever, wondering what it was thinking. My poor mom thought I was retarded.

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#109

Post by zeroonetwothree »

flatline wrote: post #59
angiolillo wrote: post #43

"The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." - Edsger Dijkstra The debate about AGI is interesting from a philosophical perspective, but from a practical perspective AI doesn't need to get anywhere close to AGI to turn the world upside down.

I don’t even know what AGI is, and neither does anyone else as far as I can tell. In the parts of the video I watched, he cites several things missing which all have to do with autonomy: continual automated updates of internal state, fully autonomous agentic behavior, etc. I feel like GPT 3 was AGI, personally. It crossed some threshold that was both real and magical, and future improvements are relying on that basic…

I think most people would consider AGI to be roughly matching that of humans in all aspects. So in that sense there’s no way that GPT3 was AGI. Of course you are free to use your own definition, I’m just reflecting what the typical view would be.

Re: Andrej Karpathy – It will take a decade to work through the issues with agents

#110

Post by fair_enough »

Imnimo wrote: post #68

>What takes the long amount of time and the way to think about it is that it’s a march of nines. Every single nine is a constant amount of work. Every single nine is the same amount of work. When you get a demo and something works 90% of the time, that’s just the first nine. Then you need the second nine, a third nine, a fourth nine, a fifth nine. While I was at Tesla for five years or so, we went through maybe three…

Reminds me of a time-honored aphorism in running:

A marathon consists of two halves: the first 20 miles, and then the last 10k (6.2mi) when you're more sore and tired than you've ever been in your life.

Post reply on HN