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

The Gentle Singularity

blog.samaltman.com

Post reply on HN
Print view View original

311–320 of 557 posts

Re: The Gentle Singularity

#311

Post by thrwwy_jhdkqsdj »

physix wrote: post #117

I started quickly reading the article without reading who actually wrote it. As I scanned over the things being said, I started to ask myself: Who wrote this? It's probably some AI proponent, someone who has a vested interest. I had to smile when I saw who it was.

I did the same thing, I thought "This post looks like a posthumous letter".

I hope LLM use will drive efforts for testing and overall quality processes up. If such thing as an AGI ever exists, we'll still need output testing.

To me it does not matter if the person doing something for you is smarter than you, if it's not well specified and tested it is as good as a guess.

Can't wait for the AI that is almost unusable for someone without a defined problem.

Re: The Gentle Singularity

#312

Post by duchef »

imgabe wrote: post #233
auggierose wrote: post #232

Earlier quoted context omitted.

So, which country is the US going to be absorbed by?

You're saying the US is simultaneously non-functional but also responsible for providing free healthcare to every other country? If we're non-functional, where is all the foreign aid providing free stuff for us? C'mon other countries, give us free stuff. It will be great for your soft power, and prevent us from doing terrorist attacks on you.

Why do you keeping making posts implying the US provides the bulk of international aid when that is demonstrably not the case (on a % of GDP basis or total dollar amount)?

Re: The Gentle Singularity

#313

Post by ixtli »

sgt101 wrote: post #296
trashtester wrote: post #287

Earlier quoted context omitted.

The "next token prediction" is a distraction. That's not where the interesting part of an AI model happens. If you think of the tokenization near the end as a serializer, something like turning an object model into json, you get a better understanding. The interesting part of a an OOP program is not in the json, but what happens in memory before the json is created. Likewise, the interesting parts of a neural net mod…

>which is in many ways similar to our subconscious thinking this is just made up. - we don't have any useful insight on human subconscious thinking. - we don't have any useful insight on the structures that support human subconscious thinking. - the mechanisms that support human cognition that we do know about are radically different from the mechanisms that current models use. For example we know that biological neu…

Yea whenever we get into this sort of “what’s happening in the network is like what’s going on in your brain” discussion people never have concrete evidence of what they’re talking about.

Re: The Gentle Singularity

#314

Post by rl3 »

imgabe wrote: post #309
rl3 wrote: post #303

Earlier quoted context omitted.

> At what speed should it be disassembled? A responsible pace that doesn't result in abrupt mass deaths due to the lack of aid continuity. > A. The US should give infinity money to everyone forever and never stop. If anyone ever dies, it's the US's fault for not supporting them enough. Nobody said that, but with operating the world's largest aid agency for the better part of a century comes massive responsibility. >…

> You're right. Hopefully those impoverished kids (many of whom are dead now) take some personal responsibility for themselves in the afterlife. To think we'd even entertain pulling their food and medicine on their schedule and not our schedule. The children? No, but their parents and the other adults running their country. That is who is responsible for providing for them. Americans have their own children they need…

>Americans have their own children they need to take care of and do not need their money seized and sent overseas to take care of other people's children.

You talk as if it's a zero-sum game, as if the two choices are mutually exclusive.

>And yes, maybe it is a "rug pull" but it was always going to be.

No reason it had to be.

Re: The Gentle Singularity

#315

Post by ixtli »

woopsn wrote: post #199
TheAceOfHearts wrote: post #150

Do you think we will get AI models capable of learning in real time, using a small number of examples similar to humans, in the next few years? This seems like a key barrier to AGI. More broadly, I wonder how many key insights he thinks are actually missing for AGI or ASI. This article suggests that we've already cleared the major hurdles, but I think there are still some major keys missing. Overall his predictions s…

Artificial intelligence is a nourished and well educated population. Plus some Adderall maybe. Those are the key insights which represent the only scientific basis for that term. The crazy thing is that a well crafted language model is great product. A man should be content to say "my company did something akin to compressing the whole internet behind a single API" and take his just rewards. Why sully that reputation…

Because they promised the latter to the investors and due to the nature of capitalism they can never admit they’ve done all they can.

Re: The Gentle Singularity

#316

Post by TheOtherHobbes »

GarnetFloride wrote: post #65
wolecki wrote: post #54

Some reasoning tokens on this post: >Intelligence too cheap to meter is well within grasp And also: >cost of intelligence should eventually converge to near the cost of electricity. Which is a meter-worthy resource. So intelligence effect on people's lives is in the order of magnitude of one second of a toaster use each day, in present value. This begs the question: what could you do with a toaster-second say 5 years…

That's what they said about electricity when nuclear power plants were proposed. What's your electricity bill like today?

Your electricity bill is set by the grift of archaic fossil energy industries. And nuclear qualifies as a fossil industry because it's still essentially digging ancient stuff out of the ground, moving it around the world, and burning it in huge dirty machines constructed at vast expense.

There are better options, and at scale they're literally capable of producing electricity that literally is too cheap to meter.

The reasons they haven't been built at scale are purely political.

Today's AI is computing's equivalent of nuclear energy - clumsy, centralised, crude, industrial, extractive, and massively overhyped and overpriced.

Real AI would be the step after that - distributed, decentralised, reliable, collaborative, free in all senses of the word.

Re: The Gentle Singularity

#318

Post by scarmig »

sgt101 wrote: post #296
trashtester wrote: post #287

Earlier quoted context omitted.

The "next token prediction" is a distraction. That's not where the interesting part of an AI model happens. If you think of the tokenization near the end as a serializer, something like turning an object model into json, you get a better understanding. The interesting part of a an OOP program is not in the json, but what happens in memory before the json is created. Likewise, the interesting parts of a neural net mod…

>which is in many ways similar to our subconscious thinking this is just made up. - we don't have any useful insight on human subconscious thinking. - we don't have any useful insight on the structures that support human subconscious thinking. - the mechanisms that support human cognition that we do know about are radically different from the mechanisms that current models use. For example we know that biological neu…

The diversity is itself indicative, though, that intelligence isn't bound to the particularities of the human nervous system. Across different animal species, nervous systems show a radical diversity. Different architectures; different or reversed neurotransmitters; entirely different neural cell biologies. It's quite possible that "neurons" evolved twice, independently. There's nothing magic about the human brain.

Most of your critique is surface level: you can add all kinds of different structural diversity to an ML model and still find learning. Transformers themselves are formally equivalent to "fast weights" (suppression and control signals). Continuous learning is an entire field of study in ML. Or, for injury, you can randomly mask out half the weights of a model, still get reasonable performance, and retrain the unmasked weights to recover much of your loss.

Obviously there are still gaps in ML architectures compared to biological brains, but there's no particular reason to believe they're fundamental to existence in silico, as opposed to myelinated bags of neurotransmitters.

Re: The Gentle Singularity

#319

Post by helloplanets »

flessner wrote: post #267

> Already we live with incredible digital intelligence, and after some initial shock, most of us are pretty used to it. Very quickly we go from being amazed that AI can generate a beautifully-written paragraph to wondering when it can generate a beautifully-written novel; It was probably around 7 years ago when I first got interested in machine learning. Back then I followed a crude YouTube tutorial which consisted o…

What's your take on Anthropic's 'Tracing the thoughts of a large language model'? [0]

> To write the second line, the model had to satisfy two constraints at the same time: the need to rhyme (with "grab it"), and the need to make sense (why did he grab the carrot?). Our guess was that Claude was writing word-by-word without much forethought until the end of the line, where it would make sure to pick a word that rhymes. We therefore expected to see a circuit with parallel paths, one for ensuring the final word made sense, and one for ensuring it rhymes.

> Instead, we found that Claude plans ahead. Before starting the second line, it began "thinking" of potential on-topic words that would rhyme with "grab it". Then, with these plans in mind, it writes a line to end with the planned word.

This is an older model (Claude 3.5 Haiku) with no test time compute.

[0]: https://www.anthropic.com/news/tracing-thoughts-language-mod...

Re: The Gentle Singularity

#320

Post by squidbeak »

AnthonyMouse wrote: post #97
GarnetFloride wrote: post #65

Earlier quoted context omitted.

That's what they said about electricity when nuclear power plants were proposed. What's your electricity bill like today?

The primary operating cost of traditional power plants is fuel, i.e. coal and natural gas. The fuel cost for nuclear power plants is negligible because the energy content is higher by more than a factor of a million. So if you build enough nuclear plants to power the grid, charging per kWh for the electricity is pointless because the marginal cost of the fuel is so low. Meanwhile the construction cost should be on pa…

Nuclear plants have other long tails costs, decommissioning and waste containment.
Post reply on HN