[HN Gopher] Behind OpenAI's plan to make A.I. flow like electricity
___________________________________________________________________
Behind OpenAI's plan to make A.I. flow like electricity
Author : typon
Score : 70 points
Date : 2024-09-26 21:24 UTC (1 days ago)
(HTM) web link (www.nytimes.com)
(TXT) w3m dump (www.nytimes.com)
| ChrisArchitect wrote:
| Gift link: https://www.nytimes.com/2024/09/25/business/openai-
| plan-elec...
| neonate wrote:
| https://archive.ph/cVled
| Ericson2314 wrote:
| > TSMC's executives found the idea so absurd that they took to
| calling Mr. Altman a "podcasting bro," one of these people said.
|
| This is very heartwarming.
| JumpCrisscross wrote:
| Is Altman another Ackman/Musk? I thought he tends to stay on
| topic when doing interviews.
| blackeyeblitzar wrote:
| Ackman and Musk stay on topic in interviews as far as I can
| tell. They may share viewpoints that much of HN disagrees
| with, but I don't see them answering questions with
| completely unrelated responses. Altman however, dodges
| questions all the time, with vague corporate speak and
| generalities. To me it looks like he is avoiding hard
| questions, although maybe he just doesn't know the answer and
| is trying to stumble his way through, or he genuinely wants
| to not give away confidential things like trade secrets.
| JumpCrisscross wrote:
| > _Ackman and Musk stay on topic in interviews as far as I
| can tell. They may share viewpoints that much of HN
| disagrees with_
|
| Makes sense. I'd consider Ackman issuing comments on
| politics off topic for a hedge fund manager. But I suppose
| he's technically on topic with a narrow view.
|
| > _Altman however, dodges questions all the time, with
| vague corporate speak and generalities_
|
| He has a documented history of dishonesty, correct?
| blackeyeblitzar wrote:
| I don't know his history fully enough to say. But it does
| seem like past claims about not getting equity or about
| keeping OpenAI's founding principles may have at least
| changed. Some claim it may be an honest change - they
| need capital to compete, attract talent, and complete
| their mission, and giving equity actually keeps Altman
| focused on OpenAI instead of side gigs and investments.
| That could be true. But I've found his evasiveness in
| interviews to be untrustworthy.
| drexlspivey wrote:
| My opinion after hearing him on multiple interviews is that
| he always just says various generalities, he can be talking
| for 5 minutes and never say anything, all noise. I don't know
| if he is under an NDA and can't say literally anything about
| OpenAI and the future or if it's just the way he talks.
| mrweasel wrote:
| Maybe he doesn't actually know anything?
| maxwell wrote:
| But he has such _force of will_.
| ben_w wrote:
| That would be my bet. He's a CEO, I've not once heard it
| suggested that he's also a researcher (though according
| to Wikipedia he's a Stanford CompSci dropout). Quite rare
| to be competent at both CEO-things _and_ the stuff you
| 're hiring people to do.
| akomtu wrote:
| Perhaps he uses an earpiece connected to an LLM.
| surfingdino wrote:
| I had the same revelation when I read the transcript of one
| of his interviews with Lex. I found it to be so free of
| detail or substance that I wanted to make sure it wasn't
| just a one case of him having a bad day, but reading
| transcripts of a couple of more interviews confirmed my
| suspicion that he has nothing of substance to say.
| shafyy wrote:
| That's all of Lex's interviews.
| Vecr wrote:
| No it's not, for example Yudkowsky gives his standard
| talk on there. I think it matters if the guest has
| something ready to go for the allotted time.
| surfingdino wrote:
| I meant Sam's answers.
| mrbungie wrote:
| That's not my impression from LeCun's or Carmack's
| (5-hours!) interviews. I would posit that, naturally, it
| depends on who is being interviewed, and that Lex also
| tries to keep the topics and their approach as accessible
| as possible.
| mrcwinn wrote:
| It may just be his public persona. I don't work with him on a
| day to day basis of course. I find him to be very
| uninteresting. He speaks with an air of aloof tech-savant but
| it feels quite hallow, faked, forced.
|
| I can't think of many examples where he said something in an
| interview and it really challenged or surprised me. By
| contrast, any random hour with Lisa Su and you will come away
| quite impressed and easily understand why she is her
| company's leader.
|
| Plenty of smart money behind Altman, though, so maybe he
| shows up stronger behind the scenes.
|
| [Edit: added "aloof" due to perceived aloofness.]
| soneca wrote:
| Magicleap also had plenty of smart money behind them. It
| ceased to be a useful signal for me then.
| botro wrote:
| "Suddenly, the chat window on Sequoia's side of the Zoom
| lights up with partners freaking out.
|
| "I LOVE THIS FOUNDER," typed one partner.
|
| "I am a 10 out of 10," pinged another.
|
| "YES!!!" exclaimed a third.
|
| What Sequoia was reacting to was the scale of SBF's
| vision....We were incredibly impressed, Bailhe says. "It
| was one of those your-hair-is-blown-back type of meetings."
|
| This is 'smart money' in reference to Sam Bankman Fried.
| benreesman wrote:
| Never having been an elite semiconductor person myself, I have
| only the dimmest intuition about the absurd science,
| technology, engineering, and mathematics that goes into it: but
| you hear these parts per trillion, measured in angstroms,
| quantum mechanics type units of account and it's hard not to
| imagine it as some world apart.
|
| I have a fond fantasy that those folks are laughing their asses
| off whenever we call the software stuff "high technology".
| petre wrote:
| They're laughing their asses off when some CEO of an AI
| company comes out and asks investors for $7T for _AI chips_.
| Sure bro, here 's 1/4 of the US GDP for your AI chips.
| a13n wrote:
| > The OpenAI chief told White House officials that A.I. data
| centers would be a catalyst for the re-industrialization of
| America, creating as many as half a million jobs
|
| Would love to hear OpenAI's explanation behind this line of
| thinking.
| tarikozket wrote:
| someone gotta plug the GPUs in and run the cables, right?
| Terr_ wrote:
| But after they plug in the GPUs and run the cables, Large
| Language Models will give everyone personal robot servants
| and jetpacks while allowing {your generous investment nation
| here} to conquer all the economies. /s
| mirekrusin wrote:
| there should be robot for that
| JumpCrisscross wrote:
| > _Would love to hear OpenAI 's explanation behind this line of
| thinking_
|
| Would have to be energy. Data centres have light human
| footprints. And Altman wants to fabricate the chips in the
| Middle East, not America [1].
|
| [1]
| https://www.bloomberg.com/news/articles/2024-04-10/openai-s-...
| segasaturn wrote:
| Yes, according to Vox Microsoft is leasing all the energy
| from the Three Mile nuclear power plant and planning to
| construct more to power their AI training. People pointed out
| how much energy Bitcoin was (and still is) wasting, AI looks
| like it's going to dwarf that amount of energy use for an
| equally questionable product.
|
| e: https://www.theverge.com/2024/9/20/24249770/microsoft-
| three-...
| soulofmischief wrote:
| "Equally questionable" is doing a lot of heavy lifting
| there. There are literally massive, paradigm-changing
| projects in both spaces _right now_ , with real users and
| communities. If you haven't gotten anything out of these
| technologies, that's on you and not them.
|
| There's all manner of insane, ambitious things being worked
| on in the open right now, if you are able to muster the
| discernment necessary to ignore the natural, unavoidable
| pile of grift that follows any new technology or hype
| cycle.
| jjulius wrote:
| >There are literally massive, paradigm-changing projects
| in both spaces _right now_ , with real users and
| communities.
|
| And in your opinion, these are...?
| soulofmischief wrote:
| Regarding crypto, IC[0] for example is an incredible
| experiment to make decentralized cloud compute viable,
| using WASM runtimes. Lots of neat crypto projects like
| this. People forget the whole point was decentralization
| and trustless systems. While the grifters settled on
| centralized systems and gaslighting users, real crypto
| projects continue to explore the possibilities of
| decentralization.
|
| With AI, I don't even know where to start. Have you used
| the internet recently? We have multimodal transformer
| models which can input and output images, video and text.
| We can create music and images using diffusion
| techniques. We have world-class text generation systems,
| which of course still need a lot of infrastructural
| support, but the viability is already being proven. We
| have cutting-edge translation tools which blow anything
| previous out the water. We have code generation and
| introspection tools, which work well enough to majorly
| augment my engineering workflow. We are developing ways
| to "speak" with your data, turning natural language into
| advance query, analysis and synthesis tasks. We have
| neural radiance fields. Lip syncing and voice translation
| tech to automate localization in powerful new ways.
|
| And we have _open models_! Anyone can dive in, fine-tune
| a model and get viable results _today_. We can already
| start creating new pipelines today, so that we aren 't
| wasting our time doing that once models get better at not
| hallucinating. The individuals and organizations who wait
| for this new technology to be perfectly reliable will get
| left in the dust by those who took the chance and did the
| pioneering work.
|
| This is still barely scratching the surface. It seems
| like there's at least one ground-breaking paper a month,
| often more than that. It's certainly easier to bash these
| technologies and communities from afar, but that attitude
| will only hinder you in the future once the tech has
| caught up. What we have already created through
| experimentation just since 2015 is stuff that I grew up
| reading about in science fiction.
|
| Could we hit an AI winter instead? Will decentralization
| ultimately be a dead end? Maybe. But it would be
| considerate to not bash those who are spending time and
| money to find out, and to not let grifters whom they
| cannot control co-opt the space and the narrative around
| it.
|
| [0] https://internetcomputer.org/
| oefnak wrote:
| > Talking with our data
|
| Beautifully put, thanks for writing my thoughts out.
| therouwboat wrote:
| Image generation with stable diffusion was useless, most
| of the time it makes images that don't even make any
| sense, like driving a car and driver is sitting in air
| outside of the car.
|
| Music generation is just mangled music from some popular
| band with custom lyrics, how can you think that you are
| creating something here?
| scotty79 wrote:
| > Data centres have light human footprints.
|
| only after they are already built
| weego wrote:
| They want huge tax credits for data centers that then won't
| employ many people or will employ visa'd migrants, but that can
| of worms can be kicked down the road to another administration.
| moogly wrote:
| All the people within creative arts will have to start working
| in the lithium mines. So that's accurate.
| throwaway4233 wrote:
| > Would love to hear OpenAI's explanation behind this line of
| thinking.
|
| My assumption is that the data centers would mostly be staffed
| by those who will have to manually audit data going in and out
| of the LLMs on a daily basis. There would also be a need to
| generate/curate the test data that the LLMs will have train on.
| There is potential for half a million jobs, but is that what
| you would want to have human effort invested in, is the real
| question.
| ben_w wrote:
| "The factory of the future is a man and a dog. The man's job is
| to feed the dog. The dog's job is to bite the man if he touches
| anything."
|
| There will therefore be a sudden supply of half a million such
| factories.
| selimthegrim wrote:
| You better tell Memphis that.
| xbar wrote:
| Altman is not a good economist.
| pwb25 wrote:
| so tired of all this AI hype and CEO stuff. just do your thing
| and let peopel buy it or not, no need to try to be the next henry
| ford or something
| elliotec wrote:
| It's all part of the Sam plan to make as much money as possible
| before whatever happens next.
| TyrianPurple wrote:
| Just pump and dump, boys. Pump and dump.
| fnordpiglet wrote:
| I'd rather we pump and dump new technologies and investment
| in renewables, scale out compute, and all the transferable
| benefits for the investment than NFTs.
|
| I'm kind of curious where it all leads to. There's a non zero
| chance it's amazing given what is in my hands today is
| already amazing compared to what was in my hands 3 years ago.
|
| I don't think the choice is like AI or curing cancer. I think
| it's more like Doge coin and meme stocks or whatever finance
| fad.
| fuzztester wrote:
| Or _pick_ the right _shovels_.
|
| _And profit!_
| vedant wrote:
| There is absolutely no way to achieve the impact of Henry Ford
| without actively trying extremely hard to be the next Henry
| Ford.
| gafferongames wrote:
| Eyes rolling so hard right now
| throwanem wrote:
| It has to be easier to pitch sovereign wealth funds on something
| that isn't even nominally nonprofit.
| kurthr wrote:
| Wait, he wants his open product to be like a utility with low
| profits for maximized reach?
|
| Or he wants his suppliers to be a regulated utility while he
| sells 100% margin products on top of it?
|
| Or he wants all the greater fools now wherever they are so he can
| get out before the collapse?
|
| It feels like this is a case of following the money to understand
| the real goals, since it's unclear to me that AGI is that goal.
| ben_w wrote:
| > it's unclear to me that AGI is that goal.
|
| Ironically, much of the observed behaviour is instrumentally
| convergent for most of the suggested ultimate goals.
|
| Trying to make a safe and aligned AGI _or_ a statutory
| government monopoly would both get about half the things we 've
| seen.
|
| The other half is stuff which is collectively mutually
| exclusive on all goals, but humans aren't perfect logical
| spheres in a vacuum, so it could still be basically any of
| them.
| krapp wrote:
| The goal is to be the man who stands at the right hand of God,
| holding the keys to the kingdom.
| riehwvfbk wrote:
| The sales pitch for the investors, of course, is that they
| will get to be God in this picture.
| krapp wrote:
| No, the AI is God. Investors just need to pay in to be one
| of the saints.
| kylehotchkiss wrote:
| yeah the whole fixation on "superintellegence" really feels
| like the a modern telling of Tower of Babel. Except this time
| we won't end up with a lot of different languages or
| whatever.
|
| at this point, the lofty goals are a distraction from
| celebrating the utility of what we have, incremental
| upgrades, and reduced resource usage.
| dogcomplex wrote:
| Yes.
|
| >Or he wants his suppliers to be a regulated utility while he
| sells 100% margin products on top of it?
|
| Near term yep - soon as regulatory capture drops. They'll make
| out like bandits.
|
| >Or he wants all the greater fools now wherever they are so he
| can get out before the collapse?
|
| Then this, by the time competition catches up anyway because
| there are no real moats here besides regulation
|
| >Wait, he wants his open product to be like a utility with low
| profits for maximized reach?
|
| Then this is what remains, in a sea of other providers.
|
| Financially? Pump and Dump, baby. Though I reckon the end
| result will still be intelligence flowing like water.
| dotancohen wrote:
| > As the availability of electricity became more widespread,
| people found better ways of using it.
|
| It's actually a nice analogy.
| mrbungie wrote:
| The level of hubris when comparing something that brings
| literal physical light to something that is not even public
| or reproducible.
| skywhopper wrote:
| All he wants is to find ways to keep propping up the
| unsustainable and insatiable beast he's bullied into existence.
| The promise of AGI is one-half threat and one-half desperate
| hope, because there's no other way to keep the bubble growing
| without government resources.
| jmakov wrote:
| There's really no way back. AI is today helping design chips
| (check Google's report on how they use it for their own designs),
| drugs etc. And having a LLM connected to a simplistic enterprise
| app is a money printing machine. Really is a new ind revolution.
| And it's held back by not enough infra and power.
| iwontberude wrote:
| I think we are seeing in realtime the resetting of expectations
| which means we don't need to "go back" rather we are realizing
| we aren't anywhere to go back from. LLMs haven't made more than
| a small dent. Non LLM inference is still more important and
| will continue to be. The problem with LLM is they input/output
| in text or images and are too general.
| tmpz22 wrote:
| > And having a LLM connected to a simplistic enterprise app is
| a money printing machine.
|
| What would be a concrete example of this? Has Salesforce
| revenue increased with their AI offerings? Has Microsofts?
| segasaturn wrote:
| This is what I was wondering reading that comment too. The
| only people I've seen making money off LLMs are spammers.
| scotty79 wrote:
| ChatGPT itself
| segasaturn wrote:
| ChatGPT loses billions of dollars, $5bn this year alone
| just on running the service. Even the subscription-based
| services like Github Copilot are losing money, $20 lost
| for every $10 subscriber according to MSFT. The only ones
| making money I can see are NVidia and maybe the oil and
| gas companies...
| jazzyjackson wrote:
| I haven't seen a single case study about copilot helping a
| company make money by rewriting their excel formulas to have
| named references with type checking to prevent clerical errors,
| which must be a billion dollar industry on its own. Why not?
| 2OEH8eoCRo0 wrote:
| Rome wasn't built in a day? Humans are also creatures of
| habit.
| scotty79 wrote:
| There isn't much evidence for benefits of strongly typed
| computer languages or object oriented programming, yet here
| we are.
| trash_cat wrote:
| I can't tell if you are being sarcastic or not.
| mrweasel wrote:
| I think the keyword here is "helping". Then you have other
| industries, creative industries, copy-writing, human resource
| and customer support where LLMs are making thing worse, way
| worse.
|
| Writing of LLMs a being useless isn't wrong and unproductive,
| but assuming that they are universally applicable, or that they
| can function without supervision is also very wrong and
| potentially dangerous.
| Jtsummers wrote:
| > Mr. Altman has since scaled his ambition down to hundreds of
| billions of dollars, the nine people said, and hatched a new
| strategy: Court U.S. government officials by first helping to
| build data centers in the United States.
|
| When all else fails, pitch your idea to the right officials in
| the USG and you'll make bank. And if you do it right, you won't
| even be doing anything at all. There are companies getting 8-10
| figures a year to fail at upgrading systems from DOS to literally
| anything not DOS. It's a very low bar, but a profitable one if
| you can pull it off.
| mesh wrote:
| Do data centers drive a lot of jobs or revenue for local or
| federal governments?
| mhh__ wrote:
| In a sense I kind of respect it but the way OpenAI are 100x-ing
| all their statements ("Universal basic compute"!) like every
| small startup has to pretend to be the next XYZ (only on a much
| bigger scale) is hilarious. Are they even in the lead at the
| moment?
| throwaway918299 wrote:
| Who is buying this absolute nonsense?! Sam Altman is the biggest
| conman of the last century.
| xbar wrote:
| He makes my list, but the list is long.
| DonHopkins wrote:
| Sam Altman will champion Direct Intelligence, while Elon Musk
| will take credit for Alternating Intelligence. Then, to prove how
| dangerous Alternating Intelligence really is, Sam will use it to
| execute dogs, fry elephants, and even immolate a few Teslas,
| before finally inventing the deadly Artificially Intelligent
| Chair to execute prisoners.
___________________________________________________________________
(page generated 2024-09-27 23:01 UTC)