TSMC execs allegedly dismissed OpenAI CEO Sam Altman as 'podcasting bro'

https://www.tomshardware.com/tech-industry/tsmc-execs-allegedly-dismissed-openai-ceo-sam-altman-as-podcasting-bro

by WithinReason • 2 years ago

485 554 2 years ago

kranke1555• 2 years ago

Might as well share the original NYT article, since this one is a poor summary of that one - https://web.archive.org/web/20240926063521/https://www.nytim...

wslh5• 2 years ago

Sorry, but I think the original article is more insightful about the complexity and reality of making chips, which TSMC knows well, compared to moonshots by people who don't fully understand the intricacies of the semiconductors supply chain.

kranke1555• 2 years ago

Huh? What original article? The Toms Hardware article is a copy and paste of the NYT one.

Rinzler893• 2 years ago

Based. The more I read about TSMC, the more I like them.

mft_3• 2 years ago

Eh; the Russians laughed at Musk, before he started SpaceX; too.

Rinzler893• 2 years ago

I doubt OpenAI will start challenging TSMC at building chips anytime soon.

malthaus• 2 years ago

indeed, that's about the most sane take on sam's plans i've seen so far from players in this craze

if this house of cards collapses, it will be magnitutes bigger than any bubble (be it hype or fraud) we've seen so far.

what a waste of resources and focus, i really can't take vc players seriously any more (from what is left after the crypto hype run)

anthomtb• 2 years ago

I have always taken both Taiwanese and TSMC culture to be fairly reserved. So if you see language like "podcasting bro" slipping out, it really means something like "fcking bullsht con artist".

jarule• 2 years ago

Taiwanese culture to be fairly reserved

That's what you see. When you're out of earshot we refer to you by several infantilizing expressions that connote npcness.

llm_trw2• 2 years ago

Taiwanese culture has also never understood software which is why they are stuck in the lower margin business of hardware.

AMD and NVidia are important now because someone else struck gold on their property. Within 5 years they won't be because both are fucking up monumentally for different reasons: AMD for being AMD [0] and NVidia for expecting 10x markup to continue indefinitely.

People hate success but LLM swarms are the next wave of improving output and they consume compute on par with training for inference.

[0] https://twitter.com/__tinygrad__/status/1839219312952766942

calf3• 2 years ago

TSMC is basically the Spacing Guild, nobody messes with them

joelthelion• 2 years ago

7 trillion is absurd. The recent advancements in AI are real, but not on the order of 7T.

aero-glide22• 2 years ago

they will be. Looking at the rate of progress (just tried o1mini), its inevitable.

jsheard5• 2 years ago

Even if we accept that premise, why should OpenAI be the ones to manage a $7T investment in hardware and datacenter development over, y'know, hardware and datacenter companies like Nvidia and Amazon? OpenAI has zero experience in those fields.

gtirloni3• 2 years ago

If it was, companies like Microsoft and Apple would have acquired them a long time ago. The fact they decided to have a partnership means they have reason to believe the hype isn't real beyond what we already know today and they don't want to have to explain it to shareholders in the near future.

AlexandrB2• 2 years ago

I once read that on the 80s it was believed that female runners would soon outpace male runners because the trend line for them was moving up so fast. This turned out not to be the case because the curve wasn't exponential but "S" shaped - the female runners eventually plateaued. It's easy to assume that exponential growth will continue indefinitely but it's rarely the case that this is true.

DebtDeflation2• 2 years ago

the rate of progress

Let's talk about that. GPT-3.5 (specifically text-davinci-003) was a massive leap over everything that came before it. GPT-4 was a massive leap over GPT-3.5. Everything since (GPT-4 Turbo, GPT-4o, GPT o1) feels like steady incremental improvement (putting aside multimodal capabilities) rather than massive leaps. I'm far from convinced that the rate of progress made with foundation models in 2022-2023 has continued in 2024 let alone from extrapolating that it will continue for the next several years.

cdchn3• 2 years ago

I think the 7T number is his concept of what he can probably get a bunch of oil sheiks to give him.

j_maffe4• 2 years ago

The GDP of UAE is 500B, 1.1T in Saudi, and 230B in Qatar. It'd take them decades to provide that kind of money.

unnouinceput• 2 years ago

I remember the story about Larry and Sergei pitching their new search engine to investors: "Will make 10B annually in maximum 10 years". Nowadays 10B is the paper clips fund at Alphabet.

anon29112• 2 years ago

You forget the money supply has been drastically inflated due to COVID, et al

KaiserPro• 2 years ago

sounds like they made a solid and accurate assessment

PedroBatista• 2 years ago

Well.. they are not wrong.

Even with the advancements made by OpenAI, some of his takes remind me of a Dollar Store Elon Musk, let alone his despicable way of doing business and seemingly complete lack of any morals.

To your average TSMC exec he's closer to a Jake Paul than to a Lisa Su or even Jensen Huang.

keepamovin• 2 years ago

There’s likely some racial bias involved, along with hesitancy to align too closely with the US, especially given the strong ties to China during these geopolitically tense times.

perfmode• 2 years ago

Only time will tell

bdndndndbve2• 2 years ago

The current AI hype wave has really hit a nerd soft spot - that we're steps away from AGI. Surely if a computer can make plausible-looking but incorrect sentences we're days away from those sentences being factually accurate! The winter after this is gonna be harsh.

llmfan2• 2 years ago

I think this is one of those "controversial" topics where we're meant to be particularly careful to make substantial comments.

bdndndndbve2• 2 years ago

I think it's substantial to say that AI is currently overhyped because it's hitting a weak spot in human cognition. We sympathize with inanimate objects. We see faces in random patterns.

If a machine spits out some plausible looking text (or some cookie-cutter code copy-pasted from Stack Overflow) the human brain is basically hardwired to go "wow this is a human friend!". The current LLM trend seems designed to capitalize on this tendency towards sympathizing.

This is the same thing that made chatbots seem amazing 30 years ago. There's a minimum amount of "humanness" you have to put in the text and then the recipient fills in the blanks.

JanSt10• 2 years ago

Using Claude 3.5 Sonnet in Cursor Composer already shows huge benefits for coding. I'm more productive than ever before. The models are still getting better and better. I'm not saying AGI is right around the corner or that we will reach it, but the benefits are undeniable. o1 added test-time compute. No need to be snarky.

jsheard5• 2 years ago

There's no accounting for taste, but keep in mind that all of these services are currently losing money, so how much would you actually be willing to pay for the service you're currently getting in order to let it break even? There was a report that Microsoft is losing $20 for every $10 spent on Copilot subscriptions, with heavy users costing them as much as $80 per month. Assuming you're one of those heavy users, would you pay >$80 a month for it?

Then there's chain-of-thought being positioned as the next big step forwards, which works by throwing more inferencing at the problem, so that cost can't be amortized over time like training can...

lynx23• 2 years ago

While I get where your fascinaton comes from...

I'm more productive than ever before.

You realize that another way to read that sentence is "I am a really bad coder".

ramblerman• 2 years ago

2 things can be true at the same time.

Op is addressing the hype that there is some linear path of improvement here and chatgpt 8.5 will be AGI.

To which people always seem to jump in with but it’s useful for me and makes me code faster. Which is fine and valid, just beside the point

aithrowawaycomm6• 2 years ago

OP could have been more substantive, but there is no contradiction between "current AI tools are sincerely useful" and "overinflated claims about the supposed intelligence of these tools will lead to an AI winter." I am quite confident both are true about LLMs.

I use Scheme a lot, but the 1970s MIT AI folks' contention that LISPs encapsulate the core of human symbolic reasoning is clearly ridiculous to 2020s readers: LISP is an excellent tool for symbolic manipulation and it has no intelligence whatsoever even compared to a jellyfish[1], since it cannot learn.

GPTs are a bit more complicated: they do learn, and transformer ANNs seem meaningfully more intelligent than jellyfish or C. elegans, which apparently lack "attention mechanisms" and, like word2vec, cannot form bidirectional associations. Yet Claude-3.5 and GPT-4o are still unable to form plans, have no notions of causality, cannot form consistent world models[2] and plainly don't understand what numbers actually mean, despite their (misleading) successes in symbolic mathematics. Mice and pigeons do have these cognitive abilities, and I don't think it's because God seeded their brains with millions of synthetic math problems.

It seems to me that transformer ANNs are, at any reasonable energy scale, much dumber than any bird or mammal, and maybe dumber than all vertebrates. There's a huge chunk we are missing. And I believe what fuels AI boom/bust cycles are claims that certain AI is almost as intelligent as a human and we just need a bit more compute and elbow grease to push us over the edge. If AI investors, researchers, and executives had a better grasp of reality - "LISP is as intelligent as a sponge", "GPT is as intelligent as a web-spinning spider, but dumber than a jumping spider" - then there would be no winter, just a realization that spring might take 100 years. Instead we see CS PhDs deluding themselves with Asimov fairy tales.

[1] Jellyfish don't have brains but their nerve nets are capable of Pavlovian conditioning - i.e., learning.

[2] I know about that Othello study. It is dishonest. Unlike those authors, when I say "world model" I mean "world."

dathinab3• 2 years ago

idk. about Claude 3.5

but if you remove implicit subventions from the AI/AGI hype then for many such tools the cost to benefit calculation of creating and operating will become ... questionable

furthermore the places where such tools tend to shine the most often places where the IT industry has somewhat failed, like unnecessary verbose and bothersome to use tools, missing tooling and troublesome code reuse (so you write the same code again and again). And this LLM based tools are not fixing the problem they just kinda hiding it. And that has me worried a bit because it makes it much much less likely for the problem to ever be fixed. Like I think there is a serious chance for this tooling causing the industry to be stuck on a quite sub-par plato for many many years.

So while they clearly help, especially if you have to reinvent the wheel for a thousands time, it's hard to look at them favorably.

PaulHoule2• 2 years ago

Usually I learn my way around the reference docs for most languages I use but CSS has about 50 documents to navigate. I’ve found Copilot does a great job with CSS questions though for Java I really do run into cases where it tells me that Optional doesn’t have a method that I know is there.

inoop• 2 years ago

"This is actually good for Bitcoin"

renegade-otter2• 2 years ago

I have yet to watch people be THAT more productive using, say, Copilot. Outside of some annoying boilerplate that I did not have to write myself, I don't know what kind of code you are writing that makes it all so much easier. This gets worse if you are using less trendy languages.

No offense, but I have only seen people who barely coded before describe being "very productive" with AI. And, sure, if you dabble, these systems will spit out scripts and simpler code for you, making you feel empowered, but they are not anywhere near being helpful with a semi-complex codebase.

rafaelmn• 2 years ago

It's a usefull coding tool - but at the same time it displays a lack of intelligence in the responses provided.

Like it will generate code like `x && Array.isarray(x)` because `x && x is something` is a common pattern I guess - but it's completely pointless in this context.

It will often do roundabout shit solutions when there's trivial stuff built into the tool/library when you ask it to solve some problem. If you're not a domain expert or search for better solutions to check it you'll often end up with slop.

And the "reasoning" feels like the most generic answers while staying on topic, like "review this code" will focus on bullshit rather than prioritizing the logic errors or clearing up underlying assumptions, etc.

That said it's pretty good at bulk editing - like when I need to refactor crufty test cases it saves a bunch of typing.

__alexs2• 2 years ago

It's not snark, it's calling out a fundamental error of extrapolating a short term change in progress to infinity.

It's like looking at the first version of an IDE that got intellisense/autocomplete and deciding that we'll be able to write entire programs by just pressing tab and enter 10,000 times.

aznumeric• 2 years ago

If you like Cursor, you should definitely check out ClaudeDev (https://github.com/saoudrizwan/claude-dev) It's been a hit in the Ai dev community and I've noticed many folks prefer it over Cursor. It's free and open-source. You use your API credits instead of subscription and it supports other LLMs like DeepSeek too.

jetsetk• 2 years ago

how come MS Teams is still trash when everyone is being so much more productive? Shouldn't MS - sitting at the source - be able to create software wonders like all the weekend warriors using AI?

layer83• 2 years ago

I’ll be waiting for these developer benefits to translate into tangible end user benefits in software.

gorjusborg• 2 years ago

Do you think AI companies will be able to afford running massive compute farms solely so coders can get suggestions?

I do not claim to know what the future holds, but I do feel the clock is ticking on the AI hype. OpenAI blew people's minds with GPTs, and people extrapolated that mind-blowing experience into a future with omniscient AI agents, but those are nowhere to be seen. If investors have AGI in mind, and it doesn't happen soon enough, I can see another winter.

Remember, the other AI winters were due to a disconnect between expectations and reality of the current tech. They also started with unbelievable optimism that ended when it became clear the expectations were not reality. The tech wasn't bad back then either, it just wasn't The General Solution people were hoping for.

gm3dmo• 2 years ago

I feel like these new tools have helped me get simple programming tasks done really quickly over the last 18 months. They seem like a faster, better and more accurate replacement for googling and Stackoverflow.

They seem very good at writing SQL for example. All the commas are in the right place and exactly the right amount of brackets square curly and round. But when they get it wrong, it really shows up the lack of intelligence. I hope the froth and bubble in the marketing of these tools matures into something with a little less hyperbole because they really are great just not intelligent.

TheCondor2• 2 years ago

It’s not snark, our industry is run on fear. If there is the tiniest flicker of potential, we will spend piles of money out of fear of being left behind. As you age, it becomes harder to deny.. 10 years ago, I was starting to believe that my kids would never learn to drive or possibly buy a car, here we are ten years later and not that much has changed, I know you can take a robotaxi in some cities but nearly all interstate trucking has someone driving.

Coding AI assistants have done some impressive things, I’ve been amazed at how they sniffed out some repetitive tasks I was hacking on and I just tab completed pages of code that was pretty much correct. There is use. I pay for the feature. I don’t know if it’s worth 35% of the world’s energy consumption and all new fabrication resources over the next handful of years being dedicated to ‘ai chips.’ We arent looking for a better 2.0, we are expecting an exponentially better “2.0” and those are very rare.

sufjkw• 2 years ago

I'm more productive than ever before.

Who are you and what are you being so productive in?

These code assistants are wholly unable to help with the day to day work I do.

Sometimes I use them to remind me what flags to use with a tarball[0], so replaced SO, but anything of consequence or creativity and they flounder.

What are you getting out of this excess productivity? A pay raise? More time with your loved ones?

[0] https://xkcd.com/1168/ (addressing the tool tip, but hilariously, in regards the comics content that would be a circumstance where I would absolutely avoid trusting one of these ‘assistants’)

freejazz• 2 years ago

Nothing snarky about pointing out AGI is nowhere near

ashkankiani• 2 years ago

LLMs make mediocre engineers into slightly less mediocre engineers, and non-engineers into below mediocre engineers. They do nothing above the median. I've tried dozens of times to use them productively.

Outside of very very short isolated template creation for some kind of basic script or poorly translating code from one language to another, they have wasted more time for me than they saved.

The area they seem to help people, including me, the most in is giving me code for something I don't have any familiarity with that seems plausible. If it's an area I've never worked in before, it could maybe be useful. Hence why the less breadth of knowledge in programming you have, the more useful it is. The problem is that you don't understand the code it produces so you have to entirely be reliant on it, and that doesn't work long term.

LLMs are not and will not be ready to replace programmers within the next few years, I guarantee it. I would bet $10k on it.

beefnugs• 2 years ago

The economics dont make sense at all:

Either you pay more and more to keep your job as it gets better, or the company pays any amount for it so they can replace you over and over as a barely useful cog.

The current state of it being cheap only exist as it is in beta and they need more info from you, the expert, until it no longer needs you

chx4• 2 years ago

George Zarkadaki: In Our Own Image (2015) describes six metaphors people have used to explain human intelligence in the last two millennia. At first it was the gods infusing us with spirit. After that it's always been engineering: after the first water clocks and the qanat hydraulics seemed a good explanation of everything. The flow of different fluids in the body, the "humors" explained physical and mental function. Later it was mechanical engineering. Some of the greatest thinkers of the 1500s and 1600s -- including Descartes and Hobbes -- assured us it was tiny machines, tiny mechanical motions. In the 1800s Hermann von Helmholtz compared the brain to the telegraph. So of course after the invention of computers came the metaphor of the brain as a computer. This became absolutely pervasive and we have a very hard time describing our thinking without falling back to this metaphor. But, of course, it's just a metaphor and much as our brain is not a tiny machine made out of gear it's also not "prima facie digital" despite that's what John von Neumann claimed in 1958. It is, indeed, quite astonishing how everyone without any shred of evidence just believes this. It's not like John von Neumann gained some sudden insight into the actual workings of the brain. Much as his forefathers he saw semblance in the perceived workings of the brain and the latest in engineering and so he stated immediately that's what it is.

Our everyday lives should make it evident how much the working of our brain doesn't resemble that of our computers. Our experiences change our brains somehow but exactly how we don't have the faintest idea about and we can re-live these experiences somewhat which creates a memory but the mechanism is by no means perfect. There's the Mandela Effect https://pubmed.ncbi.nlm.nih.gov/36219739/ and of course "tip of my tongue" where we almost remember a word and then perhaps minutes or hours later it just bursts into our consciousness. If it's a computer why is learning so hard? Read something and bam, it's written in your memory, right? Right? Instead, there's something incredibly complex going on, in 2016 an fMRI study was made among the survivors of a plane crash and large swaths of the brain lit up upon recall. https://pubmed.ncbi.nlm.nih.gov/27158567/ Our current best guess is somehow its the connections among neurons which change and some of these connections together form a memory. There are 100 trillion connections in there so we certainly have our task cut.

And so we are here where people believe they can copy human intelligence when they do not even know what they are trying to copy falling for the latest metaphor of the workings of the human brain believing it to be more than a metaphor.

imiric4• 2 years ago

That's interesting, but technology has always been about augmenting or mimicking human intelligence, though. The Turing test is literally about computers being able to mimic humans so well that real humans wouldn't be able to tell them apart. We're now past that point in some areas, but we never really prioritized thinking about what intelligence _actually_ is, and how we can best reproduce it.

At the end of the day, does it matter? If humans can be fooled by artificial intelligence in pretty much all areas, and that intelligence surpasses ours by every possible measurement, does it really matter that it's not powered by biological brains? We haven't quite reached that stage yet, but I don't think this will matter when we do.

ToValueFunfetti• 2 years ago

Helmholtz didn't say a brain was like a telegraph; he was talking about the peripheral nervous system. And he was right, signals sent from visual receptors and pain receptors are the same stuff being interpreted differently, just as telegraphing "Blue" and "Ouch" would be. That and the spirit of the gods have no place on this list and strain the argument.

Hydraulics, gear systems, and computers are all Turing complete. If you're not a dualist, you have to believe that each of these would be capable of building a brain.

The history described here is one where humans invent a superior information processor, notice that it and humans both process information, and conclude that they must be the same physically. The last step is obviously flawed, but they were hardly going to conclude that the brain processes information with electricity and neurotransmitters when the height of technology was the gear.

Nowadays, we know the physical substrate that the brain uses. We compare brains to computers even though we know there are no silicon microchips or motherboards with RAM slots involved. We do that because we figured out that it doesn't matter what a machine uses to compute; if it is Turing complete, it can compute exactly as much as any other computer, no more, no less.

ricardobayes• 2 years ago

People saying LLM to replace programming jobs is like saying blockchain is going to replace home/car titles and proof of ownership.

rubyfan• 2 years ago

Reminds me of autonomous vehicles a couple of years back. Or even AI a couple of years back, remember Watson? The hype cycle was faster to close that time.

sixQuarks2• 2 years ago

I see you haven’t tried the latest FSD build from Tesla.

autoconfig2• 2 years ago

When you do something that is extraordinarily hard, sometimes it takes longer than you expect. But now we're here: https://techcrunch.com/2024/08/20/waymo-is-now-giving-100000...

Jach2• 2 years ago

IBM Watson was more than a couple years back. The Jeopardy event was in 2011. It's currently 2024. As for cars, I don't know what you're referring to specifically, and the hype is still ongoing as far as I can tell.

It has taken 10+ years to get to present day, from the start of the "deep learning revolution" around 2010. I vaguely recall Uber promising self-driving pickups somewhere around 8-10 years ago. A main difference between current AI systems and the systems behind the cyclical hype cycles ongoing since the 1950s is that these systems are actually delivering impressive and useful results, increasingly so, to a much larger amount of people. Waymo alone services tens of thousands of autonomous rides per month (edit: see sibling comment, I was out of date, it's currently hundreds of thousands of rides per month -- but see, increasingly), and LLMs are waaaaay beyond the grandparent's flippant characterization of "plausible-looking but incorrect sentences". That's markov chains territory.

slibhb4• 2 years ago

"Plausible-looking but incorrect sentences" is cheap, reflexive cynicism. LLMs are an incredible breakthrough by any reasonable standard. The reason to be optimistic about further progress is that we've seen a massive improvement in capabilities over the past few years and that seems highly likely to continue for the next few (at least). It's not going to scale forever, but it seems pretty clear that when the dust settles we'll have LLMs significantly more powerful than the current cutting edge -- which is already useful.

Is it going to scale to "superintelligence?" Is it going to be "the last invention?" I doubt it, but it's going to be a big deal. At the very least, comparable to google search, which changed how people interact with computers/the internet.

stackghost5• 2 years ago

when the dust settles we'll have LLMs significantly more powerful than the current cutting edge -- which is already useful.

LLMs, irrespective of how powerful, are all subject to the fundamental limitation that they don't know anything. The stochastic parrot analogy remains applicable and will never be solved because of the underlying principles inherent to LLMs.

LLMs are not the pathway to AGI.

__alexs2• 2 years ago

"Plausible-looking but incorrect sentences" is cheap, reflexive cynicism. LLMs are an incredible breakthrough by any reasonable standard

No it isn't. The previous state of the art was markov chain level random gibberish generation. What OP described is an enormous step up from that.

slashdave4• 2 years ago

and that seems highly likely to continue for the next few (at least)

Why? Text training data is already exhausted.

tester7562• 2 years ago

"Plausible-looking but incorrect sentences" is cheap, reflexive cynicism.

Literally today I used Bing and it was making up API parameters.

Code example looked fine, but didnt reflect reality.

DebtDeflation2• 2 years ago

AI (or more properly, ML) is all around us and creating value everywhere. This is true whether or not we EVER reach AGI. Honestly, I stop reading/listening whenever I read/hear mention of AGI.

moron4hire2• 2 years ago

Also causing lots of harm, too, e.g. police departments using AI systems that get the wrong suspect.

amelius• 2 years ago

The winter after this is gonna be harsh.

The winter is going to be warm because of all the heat generated by GPUs ;)

wongarsu4• 2 years ago

If this winter comes, the sudden availability of cheap used enterprise GPUs is going to be a major boon for hobbyist AI training. We will all have warm homes and sky high electricity bills

yapyap2• 2 years ago

as will the summer, spring and autumn.

global warming is killing us all

klamma• 2 years ago

No but it's the first time that we have a clear picture how it could go.

And.its not just incorrect sentences it's weird questions which are getting answered a lot better than ever before.

Why are you so dismissive? Have you ever talked or wrote with a computer which felt anything like a modern LLM? I have not

AndrewKemendo• 2 years ago

Re:AGI

LLMs don’t have any ability to choose to update their policies and goals and decide on their own data acquisition tasks. That’s one of the key needs for an AGI. LLM systems just don’t do that / they are still primarily offline inference systems with mostly hand crafted data pipelines offline rlhf shaping etc…

There’s only a few companies working on on-policy RL in physical robotics. That’s the path to AGI

OpenAI is just another ad company with a really powerful platform and first mover advantage.

They are over leveraged and don’t have anywhere to go or a unique dataset.

zmgsabst2• 2 years ago

They do — but only when they’re trained on past outputs and with a willing partner.

For instance, a number of my conversations with ChatGPT contain messages it attempted to steer its own future training with (were those conversations to be included in future training).

dheera• 2 years ago

The winter after this is gonna be harsh.

That'll be investor types who bring this stupid "winter" on, because they run their lives on hype and baseless predictions.

Technology types on the other hand don't give a shit about predictions, and just keep working on interesting stuff until it happens, whether it takes 1 year or 20 years or 500 years. We don't throw a tantrum and brew up a winter storm just because shit didn't happen in the first year.

In early 2022 there was none of this ChatGPT stuff. Now, we're only 2 years later. That's not a lot of time for something already very successful. Humans have been around for several tens of thousand years. Just be patient.

If investors ran the show in the 1960s expecting to reach the moon with an 18 month runway, we'd never have reached the moon.

poopbutt102• 2 years ago

I appreciate skepticism and differing opinions, but I'm always surprised by comments like these because it's just so different from my day-to-day usage.

Like, are we using entirely different products? How are we getting such different results?

mrmetanoia• 2 years ago

I think the difference is people on HN are using these "AI" tools as coding assistance. For which, if you know what you're doing, they are pretty useful. They save trips to stack overflow or documentation diving and can spit out code that often is less time to fix/customize than it would have been to write. Cool.

A lot of the rest of the world are using it for other things. And at these other things, the results are less impressive. If you've had to correct a family member who got the wrong idea from whatever chat bot they asked, if you've ever had to point out the trash writing in an email someone just trusted AI to write on their behalf before it got sent to someone that mattered, or if you've ever just spent any amount of time on twitter with grok users, you should be exceptionally and profoundly aware of how unimpressive AI is for the rest of the world.

I feel we need less people complaining about the skepticism on HN and more people who understand these skeptics that hang out here already know how wonderful a productivity boost you're getting from the thing they're rightly skeptical about. Countering with "But my code productivity is up!" is next to useless information on this site.

egocodedinsol2• 2 years ago

this attitude is so ridiculously disingenuous. Surely if a computer can score incredibly well on math olympiad questions, among other things, "a computer can make plausible-looking but incorrect sentences" is dismissive at best.

I have no idea about AGI but honestly how can you use claude or chatgpt and come away unimpressed? It's like looking at spaceX and saying golly the space winter is going to be harsh because they haven't gotten to Mars yet.

Night_Thastus• 2 years ago

There's a big difference between those two examples.

Mars is hard but there are paths forward. More efficient engines, higher energy density fuels, lighter materials, better shielding, etc, etc. It's hard but there are paths forward to make it possible with enough time and money. We have an understanding of how to get from what we have now to what makes Mars possible.

With LLMs, there is no path from LLM -> gAI. No amount of time, money or compute will make that happen. They are fundamentally a very 'simple' tool that is only really capable of one thing - predicting text. There is no intelligence. There is no understanding. There is no creativity or problem solving or thought of any kind. They just spit out text based on weighted probabilities. If you want gAI you have to go in a completely different direction that has no relationship with LLM tools.

Don't get me wrong, the work that's been done so far took a long time and is incredibly impressive, but it's a lot more smoke and mirrors than most people realize.

sub7• 2 years ago

You are absolutely correct about where we are but don't underestimate what 100s of billions of $ can build as well. There are already credible teams working on "math AI" and "truth AI" which will likely end up combing bullshit generating LLMs with traditional but automated relational db retrievals and produce output that is both believable and correct.

IMO it will be done vertical by vertical, with no standard interface coming for a while.

dreamcompiler2• 2 years ago

One time long ago there were people living on an island who had never had contact with anybody else. They marveled at the nature around them and dreamed of harnessing it. They looked up at the moon at night and said "Some day we will go there."

But they lived in grass huts and the highest they had ever been off the ground was when they climbed a tree.

One day a genius was born on the island. She built a structure taller than the tallest tree. "I call it a stepladder," she said. The people were amazed. They climbed the stepladder and looked down upon the treetops.

The people proclaimed "All we have to do now is make this a little higher and we can reach the moon!"

tim3333• 2 years ago

Impressive stepladder.

nhnske• 2 years ago

but we reached the moon, didn't we?

HDThoreaun• 2 years ago

Really feel like this story backfires when you remember that people did go to the moon. It only took 63 years to go from the first airplane to landing on the moon.

DirkH• 2 years ago

This analogy breaks down when you consider what "attention is all you need" did for us.

A mere 5 years ago I was firmly on the camp (alongside linguists mainly) that believed intelligence requires more than just lots of data and a next token predictor. I was clearly wrong and would have lost a $1000 bet it I had put my money where my mouth was back then. Anyone not noticing how far things have come I think are mostly moving goalposts and falling to hindsight bias.

A better analogy is that the genius person in the village built a step ladder made of carbon nanotubules. Some people proclaimed "All with have to do is keep going and we can reach the moon with a very tall ladder!' Other people - many quite smart - proclaimed reasonably: "This is impossible. You are not realizing the unique challenges and materials we have not yet researched we need to build something like that."

Some in society kept building. The ladder kept getting higher. They run into issues like oxygen and balance so the ladder is redesigned into an elevator. Challenges came and were thought insurmountable until redesigns were still found to work with the miraculous carbon nanotubule material which seemed like a panacea for every construction ill.

Regardless of how high the now elevator gets and regardless of how many times the elevator gets higher than what the naysayers firmly believed is impossible the same naysayers keep saying they will never get much higher.

And higher the elevator grows.

Eventually a limit is reached, but that limit ends up being far higher than any naysayers ever thought possible. And when the limit is reached the naysayers all gathered and said "told you this would be the limit and that it was impossible!'

The naysayers failed to see the carbon nanotubules for the revolutionary potential that it had, even if they were correct that it wasn't enough.

And little did everyone know that their society was mere months ago from another genius being born that would give them another catalyst on the order of carbon nano tubules that would again lead to dramatic unexpected and long-term gains to how high the elevator can grow.

fngjdflmdflg2• 2 years ago

The winter after this is gonna be harsh.

The difference is current ML already has real use cases right now in its current form. Some examples are OCR, text to speech, speech to text, translation, recommendations (for eg. Facebook Tiktok etc.) and simple NLP tasks ("was [topic] mentioned in the following paragraph"). Even if AGI is proved impossible, these are real use cases that hold billions in value. And ML research is also considered a prestigious and interesting field academically and that will likely not change even if investors give up on funding AGI.

okdood64• 2 years ago

OCR, text to speech, speech to text, translation, recommendations

You missed the point of the parent comment's post. He's talking about the current post chatbot GenAI hype (i.e., the massive amounts of funding being poured into companies specifically after this turning point).

mensetmanusman• 2 years ago

That’s fine, the winter after this will be a productive one, because even a simulacrum of intelligence that GPT is, is useful to some extent.

sporkland• 2 years ago

I'm definitely of two minds on this topic: 1. I'm getting value out of the current batch of models in the form of accurate Q&A/summaries as well as tweaking or generating clear prose or even useful imagery well beyond what I'd ever considered possible from computers before the last two years. 2. It definitely has limits and can be a struggle to get exactly what I want and the more I try to refine something the worse it gets if the initial answer wasn't perfect.

It really feels like a substantive step forward in terms of computer utility kind of like spreadsheets, databases, apps. We'll see how far it takes us down the line of human replacement though.

tim3333• 2 years ago

120 IQ on a mensa test. https://archive.ph/OZ0sj

fennecfoxy4• 2 years ago

I don't think most people who are reasonably into AI think we're on the cusp of AGI. But I think it's made a lot of people who previously said "it will never be possible" rethink their feelings about it.

Definitely in the coming decade, we can prepare for a lot of the simpler tasks in an office to be taken over by AI. There are plenty of scenarios in which someone is managing a spreadsheet because an SME doesn't have the money to hire developers to automate & maintain that process - with advanced LLMs they can get it done by asking it to.

dmead• 2 years ago

He's a grifter. Podcast bro is charitable.

threeseed4• 2 years ago

So many weird things going on with this.

a) Given how geopolitical semiconductors and AI are right now. Why he is running around the world running his mouth seemingly without the cooperation of the US government.

b) What does Microsoft have to say about OpenAI building a parallel cloud.

c) OpenAI has zero competence in designing and building their own end to end stack but yet they are going to jump straight to $7T worth of infrastructure. Even Apple acquired PA Semiconductor and a number of smaller startups to build out their team.

thruway5162• 2 years ago

Didn't you know, according to ___ [insert favorite visionary] if you're not building an impossible moonshot you're a nothing and a nobody.

anon29112• 2 years ago

a) Given how geopolitical semiconductors and AI are right now. Why he is running around the world running his mouth seemingly without the cooperation of the US government.

As an American citizen you don't need to cooperate with the US government to talk to other people.

threeseed4• 2 years ago

Any citizen of the United States, wherever he may be, who, without authority of the United States, directly or indirectly commences or carries on any correspondence or intercourse with any foreign government or any officer or agent thereof, with intent to influence the measures or conduct of any foreign government or of any officer or agent thereof, in relation to any disputes or controversies with the United States, or to defeat the measures of the United States, shall be fined under this title or imprisoned not more than three years, or both.

https://en.wikipedia.org/wiki/Logan_Act

joeybloey3• 2 years ago

They are smart enough to know that none of this will happen, and will have cashed out long before it all crashes and burns

dboreham• 2 years ago

This guy doesn't sound like he understands engineering (checks Wikipedia...ok dropped out of Stanford). Perhaps figure out how to not need the money and electricity from the entire planet first before embarking on plan to use same.

yieldcrv• 2 years ago

a little bit of column a, a little bit of column b

factors are that international markets rely less on local cults of personality and just the financial, it’s also fair to say that TSMC is risk averse, most amusingly is that these personality cults work out fine if you throw enough money at them so I can’t fault this podcast bro for trying

isodev2• 2 years ago

I'm absolutely flabbergasted how much money and political capital is spent on vaporware, while there are very concrete and very impactful challenges we could be addressing instead.

brookst3• 2 years ago

There’s a lot of money to be made if you can reliable distinguish the two in advance.

isodev2• 2 years ago

I don't think there is appetite for this, in big tech especially. People are looking for glamorous next big things, everyone wants to be the first in some green field thing instead of digging into "the boring stuff we've been hearing about since forever". Imagine, even Apple completely enshittified their OSes just so they're on the bandwagon of AI with zero added value.

drawkward2• 2 years ago

There's even more money to be made off the backs of those who can't.

petee2• 2 years ago

My grandma loved to remind my dad about how he didn't think CDs would become a thing, and to not invest in Philips (as i recall)

And he was tech savy, building circuits, computers etc, he just didn't think it would be reliable or better than what existed... definitely a vinyl guy.

nicce• 2 years ago

Imagine how much food you get for every person in U.S. with 7 trillion. You probably get some roofs too.

layer83• 2 years ago

You won’t get those 7 trillions back though, which is what they’d be hoping for.

anon29112• 2 years ago

The US has -- if anything -- too much food. If someone is starving in America, it's 100% due to them having no interest in acquiring the free food that is widely available almost everywhere.

anon29112• 2 years ago

AI is being used for all sorts of 'very concrete' and 'very impactful' challenges, and has been for the last decade.

csomar• 2 years ago

Capital is spent to acquire more power, not to give power to some random people. This is the reality of the world we are living in.

AlexandrB2• 2 years ago

It's ironic that Sam Altman's background is in YC, because this is the opposite of startup thinking. Instead of scrappy disruption he seems to want massive investment upfront with only vague ideas of what the technology could be used for.

torlok• 2 years ago

The second sentence describes a lot of startups.

pyb• 2 years ago

It describes bad startups.

AdamN• 2 years ago

They already know that some of these upcoming model builds are going to require $100MM and possibly even a billion dollars. That is just the compute costs for building a single model. Product-market fit is basically established and the cost structure is basically understood so he needs the money for pretty straighforward reasons.

With that said it's a gamble since somebody might come along with a $10MM model based on some new technique and then OpenAI's cost structure becomes a problem. Presumably their scientists could adapt pretty quickly though if that happened.

stackghost5• 2 years ago

he seems to want massive investment upfront with only vague ideas of what the technology could be used for

That sounds exactly like venture-funded startup thinking to me.

hlanders10• 2 years ago

He may have the last laugh. His traveling around the world and threatening to build fabs in the UAE and Taiwan is a diplomatic scheme to get the U.S. hawks into action.

And lo and behold, ClopenAI has already hired CHIPS act people:

"To bolster its efforts, OpenAI has hired Chris Lehane, a Clinton White House lawyer, as its vice president of global policy, along with two people from the Commerce Department who worked on the CHIPS Act, a bipartisan law designed to increase domestic chip manufacturing. One of them will manage future infrastructure projects and policy."

We'll build plants overseas if you don't give us CHIPS Act money is a great scheme.

CharlieDigital5• 2 years ago

Build fabs

Such a misguided take.

If it were that easy to build fabs to compete with TSMC, Intel would do it. China would do it.

Hardware is way harder than software.

iphoneisbetter3• 2 years ago

Or maybe, just maybe, TSMC has gaslit all of us to think it is much harder than it really is. If Taiwan isn't needed for its chips, what is the strategic interest of the USG?

klhah• 2 years ago

Threatening to build fabs is not addressed at you but designed to trigger idiot politicians with influence to unlock CHIPS money. The politicians do not know or care if building fabs is possible.

mrweasel• 2 years ago

Which is also why it made good sense for TSMC to dismiss Altman. Some software/crypto/podcast/AI-hype bro shows up and asks for 36 fabs to be build, with no plan for monetizing them. Intel is struggling with their fab business as is, it's a difficult business to be in. Asking for 36 fabs, worth 7 trillion dollars is just absurd. Even if TSMC got the money up front, they'd still be holding those fabs and paying for their upkeep or dismantlement when/if OpenAIs predictions turn out to be wrong.

It's just an insane risk to take on.

alephnerd2• 2 years ago

We'll build plants overseas if you don't give us CHIPS Act money is a great scheme

The CHIPS act has provisions to help fund fabs in allies.

This is why Biden and Modi announced a Feb dedicated to US and Indian defense systems [0] at the QUAD summit as well as as the designation of the UAE as a "Major Defence Partner" along with India [1] which includes tech transfer conditions.

A significant portion of the CHIPS and IRA is set aside to help with subsidizing international allies tech and innovation ecosystems in order to ensure they don't lean towards China [2]

[0] https://www.bloomberg.com/news/articles/2024-09-23/biden-mod...

[1] https://www.reuters.com/world/us/harris-plans-raise-gaza-cea...

[2] https://cset.georgetown.edu/publication/agile-alliances/

Prbeek• 2 years ago

These designations are nothing but schemes to lock in major weapon buyers into the US MIC ecosystem.

actuallyalys• 2 years ago

The analogy to electricity doesn’t make sense to me. OpenAI’s and their competitors’ products are already widely available, companies have added to AI to a bunch of apps, and the cost per token has already come down significantly. (Edit: not to mention openly licensed models.)

I also don’t see think it’s historically accurate. My understanding was that widespread electrification were motivated by lighting, the radio, and other applications. Getting electricity to more people probably helped spur further development, sure, but there were already “killer apps.”

anon29112• 2 years ago

The 'token economy' is but one facet of an AI revolution. From my perspective, the main point of LLMs has been to simply be marketing. No one wants to hear about protein folding, physics sim, etc.

pphysch• 2 years ago

AI algorithms alone definitely don't represent "energy" in a meaningful way, but high-quality data is a commodity.

Part of the AI scam peeling back the facade on every AI product and realizing "oh this is actually pretty useless unless we already have high-quality automated data systems internally at $ORG", and the advancements in AI only make that nominally more accessible. No, synthetic data is not the solution, and Copilot won't build you a robust data platform.

Digitalization and digital literacy are long-term trends that AI is at the whim of, but pretends otherwise.

redleader55• 2 years ago

He seems to be the same kind of individual as Musk: able to bring together teams that build great things, but also able to bring about drama and ego. The whole musical chair act with profit/non-profit OpenAI, and getting rid of many of the folks that created the models they are famous for.

leobg4• 2 years ago

I’m sure he’d love that comparison. But Elon is responsible for PayPal, Tesla, SpaceX, OpenAI and now Grok. I have a Starlink antenna on my roof because of him, and I take my kids to school safely and quietly in a car that, 15 years ago, someone willing to spend 20 million dollars would not have been able to buy. Sam? An iris scanning crypto scam. I think that’s all he ever got off the ground. With ChatGPT, he positioned himself at the right place at the right time. And by dubious means, too.

jcranmer• 2 years ago

Elon is responsible for PayPal

Actually, he was responsible for the competitor that PayPal bought out because the market wasn't big enough for two companies, and was forced out of PayPal because he was pushing ruinously bad ideas.

FireBeyond4• 2 years ago

Bought PayPal because he couldn't do it, was the CEO for four months in early 2000 before being fired by the board the morning after he left for his honeymoon because he'd spent most of that four months trying to throw away a working Java build because he didn't understand Solaris or Java, so felt that it should be re-written in Classic ASP because he understood that.

"Responsible" indeed.

apwell234• 2 years ago

What PG see in him as a teenager to proclaim

he is 'Michael jordan of listening' or that he knew within 3 mins that sam is next bill gates.

These are actual paul graham quotes not something from Silicon Valley TV show.

Edit: to ppl responding "you should already know know paul graham is a grifter" . I don't believe is the case given his essays about meritocracy.

Invictus0• 2 years ago

Sequoia wrote 13,000 words about how Sam Bankman-Fried was god's gift to mankind. VCs are not serious.

Bankman-Fried's first pitch meeting with the VC firm's partners during its Series B round is also included in the profile, in which the founder laid out his vision to investors for FTX to become a super-app — all while he was playing the video game "League of Legends" on his computer.

leobg4• 2 years ago

I imagine that means that he observes a lot and thinks a lot before he acts. Which is great. I just wish that he had some higher God than fame and power.

snapcaster• 2 years ago

You don't see the grift here? People collude to hype eachother up to increase prestige and social status. It sounds like you take things said in public by billionaires at face value. I would suggest you stop doing that

lvl155• 2 years ago

Sam is basically a marketing person at this point. If you have the chops to be a leading AI dev, then you likely don’t need or want someone like him to skim a big chunk off of top.

causal• 2 years ago

There's a line between product hype and cult-like faith that I think OpenAI straddles way more than any startup I've seen. And it's pretty disappointing how shamelessly they milk that following, even trying to replicate "Her" to feed that particular fantasy in the fanboys.

Comparing the original mission as a nonprofit researching AI safety vs. a for-profit demanding the planet pour trillions of resources into its growth... Wild.

dizzydes5• 2 years ago

Three things that get me about current AI discourse:

- The public focus on AGI is almost a distraction. By the time we get to AGI highly-specialised models will have taken jobs from huge swaths of the population, SWE and CS are already in play.

- That AI will need to carry out every task a role does to replace it. I see this a lot on HN. What if SWEs get 50% more efficient and they fire half? That's still a gigantic economic impact. Even at the current state of the art this is plausible.

- The notion that everyone laid off above will find new employment from the opportunities AI creates. Perhaps it's just a gap in my knowledge. What opportunities are so large they'll make up for the economies we're starting to see? I understand the inverting population pyramid in the Western world helps this some also (more retirees/less workers).

schnitzelstoat• 2 years ago

I agree on the first two points.

On the third point, I think we've always seen this happen even in massive shocks like the Industrial Revolution (and the Second Industrial Revolution with assembly lines etc. and the Computer Age)

It might be hard for people to retrain to whatever the new opportunities are though. Although perhaps somewhat easier nowadays with the internet etc.

beepbooptheory3• 2 years ago

The industrial revolution didn't lay everyone off, it just made everyone's jobs worse in every possible sense.

kranke1555• 2 years ago

The Industrial Revolution was a period of massive economic growth that was coupled with decline in the quality of life of the average worker.

This Economist article talks about how looking at the historical data, historians now see that the height of the average Englishman actually went down due to malnutrition. https://web.archive.org/web/20210905065401/https://www.econo...

The myth that the Industrial Revolution was a wonderful time is just that, a myth. The actual reality of the AI revolution will likely be the same. Record number of billionaires and record number of people in deep poverty at the same time.

visarga4• 2 years ago

One thing AI is good at is explaining things, we can retrain faster and better with it.

wil421• 2 years ago

Just like robo-taxis are supposed to be driving us around or self driving cars. Not to mention the non-fiat currency everyone can easily use to buy goods nowadays.

hiddencost• 2 years ago

Waymo is doing 100k paid driverless trips a week with significantly better safety than humans in matched conditions.

dijit• 2 years ago

To be fair, I haven't touched real money in about 3 years.

If it's called "bank-coin"; "swedish crowns" or "bitcoin" it doesn't matter because it's all digital anyway.

On that count, the technological innovation is here, but it's centralised in a few "trusted" entities, just like everything.

napoleoncomplex• 2 years ago

Waymo was providing 10,000 weekly autonomous rides in August 2023, 50,000 in June 2024, and 100,000 in August 2024.

Not everything has this trajectory, and it took 10 years more than expected. But it's coming.

Not saying AI will be the same, but underestimating the impact of having certain outputs 100x cheaper, even if many times crappier seems like a losing bet, considering how the world has gone so far.

dizzydes5• 2 years ago

In the first case there are inherent safety constraints preventing it and thus its not available to public to freely use. It's highly regulated. With GPT to writing code, it is already generally available and in heavy use. There are no such life-and-death concerns in the main.

In the second case there are inherent technical challenges to using non-fiat currency and the fx volatility with fiat is wild. There are also barriers and inconveniences to conversion. With GPT writing code, the user can review for quality and still be many x more productive and there is far fewer fees and risk of loss.

It's risky to take two failed or slow innovations and assume that all innovations will be failed or slow.

visarga4• 2 years ago

What if SWEs get 50% more efficient and they fire half?

Zero sum game or fixed lump of work fallacy. Think second order effects - now that we spend less time repeating known methods, we will take on more ambitious work. Competition between companies using human + AI will raise the bar. Software has been cannibalizing itself for 60 years, with each new language and framework, and yet employment is strong.

typewithrhythm• 2 years ago

It's probably true, but just not for SWEs. Many roles will go the way of secretarys; the cost of making an administrative tool will decrease to the point where there is less need for a specialised role to handle it. The question is going to be about the pace of disruption, is there something special about these new tools?

dizzydes5• 2 years ago

Fair, there are probably two threads to this.

New product that push that bar and can command a decent margin (and good staff salaries) as long as there's a business case/demand and feature-sets that currently command a decent margin will be available for dirt cheap prices (managed by one or two person outfits).

Your comment really got me thinking, it's time to upskill haha. Aside from biotech and robotics do you see any areas particularly ripe for this push?

snowwrestler5• 2 years ago

If an AI can do my job, why would my employer fire me? Why wouldn’t they be excited to get 200% productivity out of me for the marginal cost of an AI seat license?

A lot of the predictions of job loss are predicated on an unspoken assumption that we’re sitting at “task maximum” so any increase in productivity must result in job loss. It’s only true if there is no more work to be done. But no one seems to be willing or able or even aware that they need to make that point substantively—to prove that there is no more work to be done.

Historically, humans have been absolutely terrible at predicting the types and volumes of future work. But we’ve been absolutely incredible at inventing new things to do to keep busy.

ninetyninenine5• 2 years ago

If an AI can do my job, why would my employer fire me? Why wouldn’t they be excited to get 200% productivity out of me for the marginal cost of an AI seat license?

They’d be excited at getting 100x of 100% output out an AI for 20 dollars a month and laying you off as redundant. If you aren’t scared of the potential of this technology you are lying to yourself.

dizzydes5• 2 years ago

It depends on who your employer is.

If they're high growth yes, if they're in the majority of businesses that are just trying to maximise profit with negligible or no growth then likely not.

no_op5• 2 years ago

Non-general AI won't cause mass unemployment, for the same reason previous productivity-enhancing tech hasn't. So long as humans can create valuable output machines can't, the new, higher-output economy will figure out how to employ them. Some won't even have to switch jobs, because demand for what they provide will be higher as AI tools bring down production costs. This is plausible for SWEs. Other people will end up in jobs that come into existence as a result of new tech, or that presently seem too silly to pay many people for — this, too, is consistent with historical precedent. It can result in temporary dislocation if the transition is fast enough, but things sort themselves out.

It's really only AGI, by eclipsing human capabilities across all useful work, that breaks this dynamic and creates the prospect of permanent structural unemployment.

fulafel• 2 years ago

We do have emplyoment problems arguably caused by tech, currently the bar of minimum viable productivity is higher than before in a lot of countries. In western welfare states there aren't jobs anymore for people who were doing groundskeeper ish things 50 years ago (apart from public sector subsidized employment programs).

We need to come up with ways of providing meaningful roles for the large percentage of people whose peg shape doesn't fit the median job hole.

consteval3• 2 years ago

So long as humans can create valuable output machines can't

Not all humans are going to be capable of this as time goes on. Those humans will be subjected to abject poverty.

The reality is not everyone can possess a highly skilled knowledge job. Not everyone can go to college. What of them?

jonplackett2• 2 years ago

When electricity got cheap - we use MORE electricity.

Think how many places you see shitty software currently.

My wife was just trying to use an app to book a test with the doctor - did not work at all. The staff said they know it doesn’t work. They still give out the app.

We are surrounded by awful software. There’s a lot of work to do- if it could be done cheaper. Currently only rich companies can make great software.

eitland• 2 years ago

We are surrounded by awful software. There’s a lot of work to do- if it could be done cheaper. Currently only rich companies can make great software.

Lots of the awful software is made by awfully rich companies - and lots of good software is made by bootstraped devs.

To mention some interesting examples, both Amazon and Google has gone from great to meh soon after they went from startups to entrenched market leaders.

rty32• 2 years ago

Well, that probably happens to some extent, but I am quite confident that some smaller shops will just say "Hey make an app that works 50% of the time and that's good enough." then fire half of the staff.

Oh, not just smaller shops, I have many issues with Android and other Google products -- from bugs to things that just don't work that have existed for decades, and there is no action on those over the years. Surely Google has the resources? Right? riiight?

This is a human problem, not a technology problem.

deathanatos• 2 years ago

What if SWEs get 50% more efficient and they fire half?

This is kinda ironic in a thread that's basically about the AI hype landscape, but you've just reduced the amount of SWE "power" your example organization has there by 25%.

tarsinge• 2 years ago

SWE became way more productive with Email and the Internet, did that lead to a significant part of them getting fired?

littlestymaar• 2 years ago

SWE and CS are already in play. What if SWEs get 50% more efficient and they fire half?

You know what happened last time we got 50% more efficient? It was when github and npm arrived. LLM are saving time and making us more efficient, but that's peanuts compared to the ability to just “download a lib that does X” instead of coding this shit on your own. And you know what happened after that? SWE position skyrocketed.

charlie02• 2 years ago

Buy stocks and try to own the means of production. Things are going to begin to flatten out in terms of salary or even decrease as competition increases due to productivity gains.

jp0001• 2 years ago

Well, he’s not their direct customer.

wg02• 2 years ago

OpenAI’s business model, as it exists today, doesn’t really inspire confidence, as it seems to exist on the promise of ‘jam tomorrow.’ Specifically, the firm has an income of approximately $3 billion per year, which is put in deep shade by its $7 billion annual expenditure.

So a yearly loss of 4 billion dollars or losing $10 million daily. The IPOs basically (mostly) have been operating just the way Bitcoin has been - Next bigger fool theory. That you pass on the potato to the next bigger fool till it ends to the last in the chain like Twitter who doesn't know what to do to churn out those inflated sums paid along the chain.

Not sure how many days OpenAI can afford to lose 10 million dollars daily but current macroeconomic environment, prospects aren't great.

shubhamjain• 2 years ago

Most of the expense can probably be attributed to training and operating their free-tier. They can shut off both of them any day and become insanely profitable. I don't think people realize the sheer magnitude of reaching $3B ARR in under two years. Is there an AI hype? Sure. Is it a crypto-like bubble just where everyone is just peddling BS? No, not even close.

wg02• 2 years ago

Training costing billions can't be. Biggest line item would be salaries. Probably.

Also, revenue you can't take home. Doesn't matter how many billions.

It is the profit that matters in the end.

gotaran2• 2 years ago

OpenAI has no clear moat though. If they shut off training, everyone would just move over to a competitor.

iphoneisbetter3• 2 years ago

I don't think people realize the sheer magnitude of reaching $3B ARR in under two years.

These things impress people who fail to digest business fundamentals. Anyone with a suit and a slick powerpoint preso can in theory start a money-incinerating machine. The real measurement is the profit. And before someone makes the "capture the market" comment - this isn't ZIRP anymore. Money costs fuckin money now.

zurfer• 2 years ago

Amazon was not profitable for many years, yet they are one of the most valuable companies in the world. Why? Because they kept growing.

What OpenAI is doing is "sustainable" if they keep growing like that. From summer 2023 to summer 2024 they 6x their revenue to 3.4B USD. If they 4x next year and 3x the year after, they would make 40B in revenue.

mrbungie2• 2 years ago

That's a common misconception: They reinvested their money back into Amazon, which when justified with a sound strategy to investors makes sense. It was kept by design at an almost 0 profit, but non profitable != operating at a loss (vs OAI, that has been predicted to loss up to 5B this year).

j_maffe4• 2 years ago

You're assuming exponential growth based on early growth rates. Those rates are extremely unlikely.

nojvek• 2 years ago

Where are you getting $3B ARR number? curious about source.

gepardi• 2 years ago

That’s hilarious. Maybe there’ll be an episode in some show a decade from now, inspired by this.

5cott0• 2 years ago

more like “anime pfp anon”

SavageBeast2• 2 years ago

I was having some kind of issue the other day - browser cache corruption issue or ChatGPT hiccuped for a few minutes - but I needed to know an answer for a thing RIGHT NOW. I reluctantly went back to Google and searched for it the old way.

Just ... OMFG ... I used to do this all the time? This used to be how you learned things? You mean I get to read 10 sites and 25 dumb answers to get to a possibly correct one? And all the while I got ads coming at me for this and that. A friend was trying to decode some chat convo we had about GLP-1 drugs and get factual information.

Google search was nothing but direct ads selling "IF YOU LIKE OZEMPIC YOU WILL LOOOOOVE NOZEMPIC - ITS ALL NATURAL!!" and the content-mill blogs and such that tout the benefits of some "NOZEMPIC" variant.

AI has changed the way I learn and research. I have a very capable tutor on demand for any subject there is, any time I want to use it. Its mostly right or "good enough for government work" as the old guys say.

I was once an AI hater - now Im an AI evangelist and Im turning people in my sphere of influence into AI users too. Once you show them the value prop they totally get it. Learning and such no longer requires manual work for the 90% case.

meiraleal2• 2 years ago

Just ... OMFG ... I used to do this all the time? This used to be how you learned things?

No. Google is really not useful anymore, it wasn't like that. It is IMPOSSIBLE to search for what you want, Google just think they know what you want and literally change the words to match. It is crazy.

SavageBeast2• 2 years ago

I think Google, with regard to Search here, is in the Skin The Sheep phase of its story arc. Innovation is done, market dominance is here, no point in innovating because they're making so much money. The total shit state that is Google Search these days is a tacit admission that search as we know it is over. No more innovating, instead lets suck as much money out of this thing as we can today because its going away. Good news is things become pure profit for Google while they wait for the inevitable changing of the guard and their replacement becomes the new normal.

finikytou2• 2 years ago

dont forget google wasnt like this at the beginning. OpenAI or competition WILL incorporate ADS into the equation and at the end we will have moved away from google to reproduce the same model

bob10293• 2 years ago

It is clear to me that Sam has never set foot inside of a semiconductor manufacturing facility. Or, if he did he definitely wasn't paying attention to what was going on around him. I don't know how you could witness these things and then make glib statements about building 36 of them.

It will likely take well over a decade to recoup investment on any new chip fab at this point. Chasing ~4 customers on one narrow use case is nonsensical from the perspective of anyone running these companies.

H8crilA• 2 years ago

It is reasonable from his point of view - you guys sink the capital, I will have more compute. And if it doesn't work out and orders stop coming in then well that's not my problem, is it?

bob10293• 2 years ago

It is reasonable from his point of view

If he wants to actually accomplish his objectives, he needs to get inside the minds of the executives that run these companies. Empathize with their concerns and then develop a strategy for walking them towards a path to build even one additional factory.

Throwing out a cartoonish figure and then hoping to be taken seriously is not something I'd expect from the CEO of something so adjacent.

joeybloey3• 2 years ago

It sounds reasonable for him because he's a podcasting bro, not a serious engineer or executive

jhanschoo• 2 years ago

It's entertaining when one sees it as sama trying to save OpenAI from massive costs by persuading governments to invest an order more magnitude into compute to artificially distort the actual costs of compute to make OpenAI's costs make sense.

no_op5• 2 years ago

In his recent "Intelligence Age" post, Altman says superintelligence may be only a few thousand days out. This might, of course, be wrong, but skyrocketing demand for chips is a straightforward consequence of taking it seriously.

rsynnott• 2 years ago

may be only a few thousand days out

This is actually quite clever phrasing. "A few thousand days" is about ten years, assuming normal usage of 'few' (ie usually a number between 3 and 6 inclusive).

Now, if you, as a tech company, say "X is ten years away", anyone who has been around for a while will entirely disregard your claim, because forward-looking statements in that range by tech companies are _always_ wrong; it's pretty much a cliche. But phrasing as a few thousand days may get past some peoples' defences.

slashdave4• 2 years ago

Only if you think scaling is the solution to AGI, which it almost certainly is not

dizzydes5• 2 years ago

Facilities aside, I hear talent is a massive bottleneck. Especially in the US.

bob10293• 2 years ago

Skilled labor is the ultimate constraint.

It's really hard to keep good employees working at these places for a long duration. Innovation is something your customers do. Your job is to make it as boring as fucking possible.

I could only stand it for 3 years before I had to quit (Samsung). I know others who still enjoy the experience though.

fsefkjln• 2 years ago

Can anyone provide a time where Sam did not come off as a slimy imposter instead of the tech visionary he would like us all to believe? Literally everything I have read on him dating back years makes him seem like a clueless buffoon who is only in it for the money. Being silver tongued, connected and cunning can make you a lot more money a lot easier than being competent and honest.

imaginationra• 2 years ago

Grifters gonna grift-

My nerd hype died as I used gpt4+ and Claude 3.5+ whatever for coding tasks and realized if a human somewhere in the training data hadn't already solved a programming problem/wrote code for a solution the LLM's are useless for said problem.

They're useful as an assistant/junior coder to do repetitive tasks/things you/other humans already have already done tho. BUT that's not going to revolutionize programming as you already have to know how to code to use them.

* and "hallucinations" should be referred to as "bullshitting" when it comes to LLM's as that's what it really is.

Mistletoe• 2 years ago

I was watching Alien last night and as they talked with Mother it was refreshing to see the ship computer just say “I don’t know” when it didn’t know something. If we could get our LLMs there it would be a nice step forward. You can definitely tell ours were made in Silicon Valley, home of the bullshit and fake it till you make it.

prewett• 2 years ago

I tried asking Gemini to summarize a non-existent poem, and it said it would need the actual source of the poem to summarize it (but provided four potential themes based on the title). I then suggested that it was a standard English course text and it apologetically said that it couldn't find any record of it, but perhaps I was thinking of this other poem? It is perhaps too helpful at coming up with a workable solution to my request, but it did seem to know when it didn't know. Although perhaps that isn't the LLM but some additional information from a search query being fed into the LLM.

anon29112• 2 years ago

My nerd hype died as I used gpt4+ and Claude 3.5+ whatever for coding tasks and realized if a human somewhere in the training data hadn't already solved a programming problem/wrote code for a solution the LLM's are useless for said problem.

See my experience is the opposite, and gpt3 was able to answer questions about the niche Haskell library I wrote better than I was. Unless you posit that there's some secret cabal of people using this Haskell library and writing answers for GPT to consume somewhere in the depths of OpenAIs org structure, it's pretty obvious to me that LLMs are quite capable.

drawkward2• 2 years ago

More cryptobro, imo

andy_ppp2• 2 years ago

Arguably compared to the technical people who do the actual work isn’t this a fairly perceptive description?

joeybloey3• 2 years ago

There's lots of smart serious people who aren't technical but the problem with Altman is that he isn't serious at all, he's a pure grifter, focused only on headlines

mpalmer• 2 years ago

This jumped out at me:

"The NYT claims it discussed OpenAI's negotiations with nine people close to the discussions but who wish to remain anonymous."

"Claims"? Generally newspapers do not lie about the sourcing for a story.

dylan6042• 2 years ago

You have qualms with "claims", yet you use "generally" to describe newspaper dogma. So you accept that newspapers do not always not lie, yet you have issues accepting that sometimes sources have not been verified?

jpm_sd• 2 years ago

At this point it's become clear that Sam Altman is an Emissary of the Eschaton (in the Strossian sense)

finikytou2• 2 years ago

they are right tho. He is a snake oil vendor and all his A team leaving due likely to his moves and unethical steps taken from day 1 are proof that TSMC is a solid business

zooq_ai2• 2 years ago

Remember when the Russians mocked Elon Musk for his rocket purchase?

Remember when Blockbuster mocked Netflix?

Remember when Steve Ballmer mocked the iPhone?

Remember when Zuck was mocked for the Metaverse?

I'm pretty sure Sam would have heard similar mockings like "You can't take on Google behemoth, podcaster bro" before.

But the sad part is, the newer tech generation is so brainwashed that it takes the side of the mocker instead of the mockee.

Watch Sam partner with some other fab like Intel and make TSMC irrelevant. Of course HN will then side with Sam

dylan6042• 2 years ago

Remember when Zuck was mocked for the Metaverse?

One of these things is not like the others. You listed 3 things proven silly with hindsight, then you post one comment that is mere hopeful wishing.

koolala• 2 years ago

If Meta makes those new AR glasses real it makes up for the Metaverse branding.

zooq_ai2• 2 years ago

Like I said, if you didn't follow Orion's demo and still parroting "Metaverse Bad" you belong in the exact same category as Blockbuster

tempodox• 2 years ago

Altman is used to deal with people whose medium is mostly hot air. I'd expect a chip manufacturer to have a healthy allergy against that. Having your process filled with 99% bullshit might be good enough if the desired outcome is investment, but if you want to produce functional hardware, you can't afford that.

asn1parse• 2 years ago

lolololololololololololololol

JCharante• 2 years ago

Even implementing a fraction of the OpenAI CEO’s ideas would be incredibly risky, the execs are said to have openly pondered.

I don’t want to live in a world where nobody takes risks.

However, the latest OpenAI statements have rolled back such talk to "mere" hundreds of billions. It is reported that years of construction time would also be needed to satisfy the OpenAI compute scaling plans.

Yeah so what? It will take years to build so build 36 mega data-centers in parallel.

the firm has an income of approximately $3 billion per year, which is put in deep shade by its $7 billion annual expenditure.

Those are tiny numbers. Airpods are a $14B/year line of products. OpenAI needs to be ready to spend 100B/year.

Likewise, Apple launched its iPhone 16 and 16 Pro earlier in the month with a lot of talk about Apple Intelligence, but the first of these AI features won’t be available on the new devices until next month.

I’m on a 15 pro max with the 18.1 PUBLIC Beta. I have these features and so do 16 users. Mark Tyson. This is terrible lazy journalism. You add half true statements to make your point.

lizknope2• 2 years ago

I read the NY Times article that someone else linked to in the thread.

I've been in semiconductor / chip design for almost 30 years at 8 different companies including 2 that owned their own fabs. I've also worked on 2 different AI accelerator chips.

Altman sounds like someone who has no idea about the industry. The numbers in the article are laughable.

This quote summed it up.

It is still unclear how all this would work. OpenAI has tried to assemble a loose federation of companies, including data center builders like Microsoft as well as investors and chipmakers. But the particulars of who would pay the money, who would get it and what they would even build are hazy.

He's trying to get a bunch of diverse companies including Microsoft to fund this but it's unclear what kind of chips he actually wants. Many of the big companies including Microsoft are designing their own custom AI chips.

But the article mentions Nvidia. Does OpenAI have any plans to design their own chips or do they just use Nvidia? It may be difficult to get Nvidia competitors to want to join his effort.

TSMC makes semiconductors for Nvidia, the leading developer of A.I. chips. The plan would allow Nvidia to churn out more chips. OpenAI and other companies would use those chips in more A.I. data centers.

calf3• 2 years ago

He's a Steve Jobs type C-level manager. The actual knowledge people are scientists such as Ilya Sutskever et al., many of whom left their company over the past year.

The question is whether LLMs are going to plateau. Either OpenAI and the tech industry system can keep progress going, or else more basic science is needed, in which case a government-project level of coordination is needed instead.

ksec2• 2 years ago

He's a Steve Jobs type C-level manager.

Steve Jobs actually knows a lot more about many things than most people realise.

ksec2• 2 years ago

Altman sounds like someone who has no idea about the industry.

I think it also sums up a lot about what I am seeing from software developers in general. Very little understanding of hardware. And even more so about hardware industry itself.

HN itself is a very good example and evidence.

lizknope2• 2 years ago

I see people on HN approach their lack of knowledge with a little more humility than the way Altman sounds in the article. He seems to be one of those people who thinks "I'm an expert in my area therefore I am an expert in all areas." The really smart people know what they know and also what they don't know and surround themselves with people that have the knowledge they lack. Someone he trusts should be telling him his ideas for fab expansion are crazy.

BaculumMeumEst2• 2 years ago

This is like posting a reaction video. Just post the actual source.

spamizbad• 2 years ago

TSMC shade aside, I think Altman is keenly aware that AI is still too capital intensive for short-to-medium term viability and I think his hope to was to probe the semiconductor industry's willingness to build-out more fabs to drive the costs down substantially. I suspect Altman was being bluntly honest with them about the economics of AI rather than enthusiastically telling them they should immediately build 37 fabs because surely that's super easy.

Blackwell cabinets are $3M each and have significant power requirements and thus you'd likely need to be selling your products at $100/month/seat to turn a slim profit.

ChrisArchitect• 2 years ago

Gift link of the NYT source: https://www.nytimes.com/2024/09/25/business/openai-plan-elec...

throw48472852• 2 years ago

This is going to sound nuts, but I think CEOs like Altman have more in common with historical monarchs than modern Democratic leaders. In a Democracy, there's more transparency into the degree to which everything the President/PM does is actually delegated out. Joe Biden may be the Commander in Chief, but just watch a political military thriller and you'll see that culturally, we understand the role of the Joint Chiefs of Staff in having the actual depth of knowledge the President lacks.

But CEOs often practice a form of charismatic leadership whereby you have to project having more knowledge than you actually do. That is because their leadership is both less democratic and less secure than their counterparts. You've got to be constantly maneuvering against challenges to your legitimacy.

I think it has something to do with the Gemeinschaft–Gesellschaft dichotomy from German sociology. You might think that a tech company would be the quintessential form of Gesellschaft, but in reality, the community of tech CEOs and VCs is far more like a social club than it is like a rational, contractual relationship.

MDJMediaLab• 2 years ago

I'd like to subscribe to your newsletter.

throw48472852• 2 years ago

I'm truly flattered, but you would just get regurgitated ideas from smarter newsletters.

thebigspacefuck• 2 years ago

Maybe the board was right to kick him out

yapyap2• 2 years ago

36 new plants, 7 trillion dollars and that would all have to be backed by AI? yikes.

I know Sam is gonna be a billionaire assetwise soon but I wonder if it’ll make him more realistic or even more “idealistic” (bigger bullshitter)

churchill2• 2 years ago

On some level, I still see Sam Altman as a visionary, but the moment that got me rolling my eyes was when we traveled to Middle East, reportedly looking to raise up to $7 trillion for AI chips.

Like, bro, all the sovereign wealth funds of the Arab Gulf hold less than $3 trillion, mostly in illiquid assets that you could take a 50% haircut if you need to liquidate. And after the $100b Vision Fund fiasco where Saudi Arabia was subsidizing "Uber for Dog-walking" startups for years (no thanks to Masayoshi Son), I'm not sure they're going to be keen subsidizing yet another bubble.

But, isn't that the default in VC? Aggressive, toxic optimism because it's not your money and you need to promise otherworldly yield to get capital.

davidivadavid• 2 years ago

Can someone enlighten me on how Sam Altman is a "visionary" of any kind? The extent of his vision, to the best of my knowledge so far, is "things will get cheap then cool stuff will happen." That's the laziest "vision" I've ever seen in my life. It depicts nothing concrete whatsoever.

stonethrowaway• 2 years ago

How do you anticipate sama will take being rejected by TSMC?

Ninjak8051• 2 years ago

Maybe Altman has no idea about the realities of semiconductor manufacturing, or maybe he does. Sometimes it takes an outsider to shake things up and produce the impossible. Probably some Verizon execs called Steve Jobs a "computer bro" when he walked in with a pre-production iPhone and asked for "half".

m3kw9• 2 years ago

This is what he needs, someone is there to tell him if it can be done or not.

htk• 2 years ago

Very biased article, with the author injecting a lot of his personal opinions on the matter.

dang• 2 years ago

I've re-upped the original article here:

Behind OpenAI's plan to make A.I. flow like electricity - https://news.ycombinator.com/item?id=41663562

Submitters: "Please submit the original source. If a post reports on something found on another site, submit the latter." - https://news.ycombinator.com/newsguidelines.html

rldjbpin• 2 years ago

as someone working in AI, if i had access to 7T USD, i would never put it all in AI research, forget in one moonshot alone to put it best.

in retrospect, press media was much more critical of meta and the whole metaverse schtick. it took long but all the money poured on the research has brought more meaningful results with the orion prototype.

we are giving this AI company too much air time. let them do their bidding by themselves.

TSMC execs allegedly dismissed OpenAI CEO Sam Altman as 'podcasting bro'

Related Stories