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

Abundant Intelligence

blog.samaltman.com

Post reply on HN
Print view View original

121–130 of 136 posts

Re: Abundant Intelligence

#121

Post by beeflet »

MountDoom wrote: post #88
beeflet wrote: post #49

Earlier quoted context omitted.

Maybe AI will become more ubiquitous. But I predict LLMs will be capped by the amount of training data present in the wild.

Ubiquity doesn't depend on the AI getting much better as much as it depends on the computational cost going down (i.e., better hardware + software optimizations). When you can put a ChatGPT-class model locally on every desktop or phone, people will use it even if the accuracy or safety isn't quite there. Just look at how people are using Grok on Twitter, or how they're pasting ChatGPT output to win online arguments,…

>When you can put a ChatGPT-class model locally on every desktop or phone, people will use it even if the accuracy or safety isn't quite there.

Will they though?

>Just look at how people are using Grok on Twitter, or how they're pasting ChatGPT output to win online arguments, or how they're trusting Google AI snippets. This is only gonna escalate.

But will they though?

Re: Abundant Intelligence

#122

Post by rooftopzen »

j_timberlake wrote: post #116
ahmedtd wrote: post #115

Earlier quoted context omitted.

Sam Altman skipped any attempt to prove his own statements right, so...

So be better than him..?

Sam Altman has been a joke for awhile now, heard only his investors defend him for their next round increase - is that who you are?

Re: Abundant Intelligence

#123

Post by rooftopzen »

tensor wrote: post #81
Sivart13 wrote: post #52

> Maybe with 10 gigawatts of compute, AI can figure out how to cure cancer. Something I've never understood: why do AGI perverts think that a superintelligence is any more likely to "cure cancer" than "create unstoppable super-cancer"

AI will do neither of those things because curing or creating cancer requires physical experiments and trials on real people or animals, as does all science outside of computer science (which is often more math than science). I can see AI being helpful in generating hypothesis, or potential compounds to synthesize, or helping with literature search, but science is a physical process. You don't generally do science ju…

Exactly - AI allows for intersections in concepts from training data; up to the user to make sense of it. Thanks for stating this (I end up repeating same thing in every conversation, but is common sense).

Re: Abundant Intelligence

#124

Post by rooftopzen »

ozten wrote: post #85
tensor wrote: post #81

Earlier quoted context omitted.

AI will do neither of those things because curing or creating cancer requires physical experiments and trials on real people or animals, as does all science outside of computer science (which is often more math than science). I can see AI being helpful in generating hypothesis, or potential compounds to synthesize, or helping with literature search, but science is a physical process. You don't generally do science ju…

There are a few fully automated wet labs and many semi-autonomous. They are called "Cloud Labs", and they will only become more plentiful. AI can identify and execute the physical experiments after using simulations to filter and score the candidate hypotheses.

Sorry but your concept of AI is marketing driven. It's probabilistic, understanding is past your pay grade.

Re: Abundant Intelligence

#125

Post by TeMPOraL »

floren wrote: post #39
TeMPOraL wrote: post #29

Earlier quoted context omitted.

That's funny, but unless LLM bugs break foundational ML codebases beyond human repair (and somehow also delete all existing code, research, and researchers), the models will likely just get better than people at this in a couple years. I mean, the trajectory so far is obvious.

disco-stu-pointing-at-chart.gif

Go short some stocks if you're so certain the trend will change any minute now (like it was always about to for the past two years).

Re: Abundant Intelligence

#126

Post by SMAAART »

Can someone help me make some sense of the 10 gigawatts of compute?

Colossus data has 230K GPUs (150,000 H100 GPUs, 50,000 H200 GPUs and 30,000 GB200 GPUs) [source https://x.ai/colossus]

Energy usage: up to 150 megawatts of electricity per day [source: https://en.wikipedia.org/wiki/Colossus_(supercomputer)]

So, when SamA talks about 10 gigawatts of compute does he mean per day or GWH (Gigawatts-hour)?

Re: Abundant Intelligence

#127

Post by tensor »

rooftopzen wrote: post #124
ozten wrote: post #85

Earlier quoted context omitted.

There are a few fully automated wet labs and many semi-autonomous. They are called "Cloud Labs", and they will only become more plentiful. AI can identify and execute the physical experiments after using simulations to filter and score the candidate hypotheses.

Sorry but your concept of AI is marketing driven. It's probabilistic, understanding is past your pay grade.

They're actually right in that there are several attempts to create automated labs to speed up the physical part. But in reality there are only a handful and they are very very narrowly scoped.

But yes, potentially in some narrow domains this will be possible, but it still only automates a part of the whole process when it comes to drugs. How a drug operates on a molecular test chip is often very different than how it works in the body.

Re: Abundant Intelligence

#128

Post by rsfern »

SMAAART wrote: post #126

Can someone help me make some sense of the 10 gigawatts of compute ? Colossus data has 230K GPUs (150,000 H100 GPUs, 50,000 H200 GPUs and 30,000 GB200 GPUs) [source https://x.ai/colossus ] Energy usage: up to 150 megawatts of electricity per day [source: https://en.wikipedia.org/wiki/Colossus_(supercomputer) ] So, when SamA talks about 10 gigawatts of compute does he mean per day or GWH (Gigawatts-hour)?

I think it’s just scale to the moon rhetoric, like “what if we used 100x more compute?”. Since the units are power and not energy, I’m going with 10 GW continuous load (for training? inference?) but I think it’s not exactly meant literally

Re: Abundant Intelligence

#129

Post by sensanaty »

toddmorey wrote: post #67
mcpar-land wrote: post #28

> As AI gets smarter, access to AI will be a fundamental driver of the economy, and maybe eventually something we consider a fundamental human right. My product is going to be the fundamental driver of the economy. Even a human right! > Maybe with 10 gigawatts of compute, AI can figure out how to cure cancer. How? > We are particularly excited to build a lot of this in the US; right now, other countries are building…

Not a word or whisper about environmental impact, either. I mean at least do some hand waving or something. I feel like a habitable planet is a fundamental right.

Duh, you just ask the magical AI "how fix global warming" and it fixes global warming, same way it'll cure cancer magically.

Re: Abundant Intelligence

#130

Post by floren »

TeMPOraL wrote: post #125
floren wrote: post #39

Earlier quoted context omitted.

disco-stu-pointing-at-chart.gif

Go short some stocks if you're so certain the trend will change any minute now (like it was always about to for the past two years).

Unfortunately as they say, the market can remain irrational longer than I can remain solvent.
Post reply on HN