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

131–136 of 136 posts

Re: Abundant Intelligence

#131

Post by Balgair »

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)?

Video. It's for video.

We don't have the compute to do video on demand right now like we do images or text or audio.

Combining all the modalities together, smoothly, at speed, and for cheap, is going to take a hell of a lot of thinking sand powered by magic rocks.

Re: Abundant Intelligence

#132

Post by heyitsguay »

Workaccount2 wrote: post #117
thrance wrote: post #37

Earlier quoted context omitted.

Stop thinking and give them money. But for real, the leap from GPT4 to GPT5 was nowhere near as impressive as from GTP3 to GPT4. They'll have to do a lot more to give any weight to their usual marketing ultra-hype.

I think people have a lot of rosy glasses and fondness for those early days, combined with general usability benchmarks being mostly saturated now. GPT-3.5 would say Dallas was the capital of USA, but GPT-4 got it every time! GPT-4 launched with 8k context. It hallucinated regularly. It was slow. One-shotting code was unheard of, you had to iterate and iterate. It fell over even doing basic math problems. GPT-5 think…

What problems?

Re: Abundant Intelligence

#133

Post by vivzkestrel »

thoughtpeddler wrote: post #99
vivzkestrel wrote: post #94

I have stopped reading all the sam altman posts. Not because I am a hater or anything. Maybe AI is a bubble, maybe it is not but one thing is clear. Your ability to do something is enforced in neural pathways that you exercise on a regular basis. If you let LLMs and AI do your architecture, database, coding, testing and just about all the things you are dreaming, then gradually but surely, your brain will lose these…

I hear you, but I don't understand the logic. If you were wealthy enough to have an executive assistant or a chief of staff who handled most of your email and admin on your behalf, is your instinct to say, "I will still write my own emails because I don't want that skill to atrophy?" No. It's simply not high-value work. I'd assume you rather do something else. You don't see the ultra-wealthy say "Oh no! Not my abilit…

except you got one part of that wrong. Most beginners are trying to generate entire UI, API, design, database schemas, and god knows what else with AI. the concept of the so called marginal AI user simply doesn't exist and you know it

Re: Abundant Intelligence

#134

Post by thoughtpeddler »

vivzkestrel wrote: post #133
thoughtpeddler wrote: post #99

Earlier quoted context omitted.

I hear you, but I don't understand the logic. If you were wealthy enough to have an executive assistant or a chief of staff who handled most of your email and admin on your behalf, is your instinct to say, "I will still write my own emails because I don't want that skill to atrophy?" No. It's simply not high-value work. I'd assume you rather do something else. You don't see the ultra-wealthy say "Oh no! Not my abilit…

except you got one part of that wrong. Most beginners are trying to generate entire UI, API, design, database schemas, and god knows what else with AI. the concept of the so called marginal AI user simply doesn't exist and you know it

I think that's a passing artifact of the current phase of AI development we find ourselves in. Capabilities (note how I'm not saying 'model intelligence', as I think various agentic flows and robust scaffolding/harnesses can lead to capabilities growth that goes beyond model intelligence plateauing) will continue to improve such that 'beginners' will gain greater and greater leverage.

If by 'marginal AI user' you mean a user who leverages AI tools to enhance the marginal utility of their labor or tasks (by making them more productive or efficient, broadly defined), then I do think that user archetype definitely exists.

Re: Abundant Intelligence

#136

Post by johneth »

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..?

It's not worth anyone's time to meticulously fact check known (and I'm being kind here) 'exaggerator' Sam Altman, because by the time you're done, he's already spread 10 more 'exaggerations'.
Post reply on HN