New "investors" are Microsoft and Nvidia. Nvidia will get the money back as revenue and fuel the hype for other customers. Microsoft will probably pay in Azure credits.
If OpenAI does not make profit within two years, the "investment" will turn into a loan, which probably means bankruptcy. But at that stage all parties have already got what they wanted.
I don’t believe this is accurate. I think this is what you’re referring to?:
Under the terms of the new investment round, OpenAI has two years to transform into a for-profit business or its funding will convert into debt, according to documents reviewed by The Times.
That just means investors want the business to be converted from a nonprofit entity into a regular for-profit entity. Not that they need to make a profit in 2 years, which is not typically an expectation for a company still trying to grow and capture market share.
Source: https://www.nytimes.com/2024/10/02/technology/openai-valuati...
At about the ten year mark, there has to be a changing of the guards from the foot soldiers who give their all that an unlikely institution could come to exist in the world at scale to people concerned more with stabilizing that institution and ensuring its continuity. In almost every company that has reached such scale in the last decade, this has often meant a transition from an executive team formed of early employees to a more senior C-team from elsewhere with a different skillset. In a world context where the largest companies are more likely to stay private than IPO, it's a profoundly important move to allow some liquidity for longterm employees, who otherwise might be forced to stay working at the company long past physical burnout.
Which world is this?
Amazon founded in 1994
1997 listed at $400m
Google founded in 1998
2004 listed at $23bn
Spotify founded in 2006
2018 listed at $27bn
Airbnb founded in 2008
2020 listed at $47bn
Epic games founded in 1991
2022 unlisted value $32bn
Space Exploration founded in 2002
2022 unlisted value $125bn
ByteDance founded in 2012
2022 unlisted value $360bnI doubt any competitor to the largest businesses in US//Europe that is actually putting up good audited numbers is staying private. Even Stripe has been trying to go public, but it doesn’t have the numbers for the owners to want to yet.
Their kind of product development [1] needs a long term thinking which public markets will not not support well
[1] ignore all the mars noise, just consider reusable i.e. cheap rockets and starlink
The long-term problem may be access to quality/human-created training data. Especially if the ones that control that data have AI plans of their own. Even then I could see OpenAI providing service to many of them rather than each of them creating their own models.
I never use the $20 plan but I access everything via API and i spend a couple of dollars per month.
Although lately I have a home server that can do llama 3.1 8b uncensored and that actually works amazingly well.
It works ok but with a large context it can still run out of memory and also gets a lot slower. With small context it's super snappy and surprisingly good. What it is bad at are facts/knowledge but this is not something a LLM is meant to do anyway. OpenWebUI has really good search engine integration which makes it work like perplexity does. That's a better option for knowledge usecases.
The only thing I see is that it hallucinates a lot when you ask it for knowledge. Which makes sense because 8B is just not a lot to keep detailed information around. But the ability to recite training knowledge is really a misuse of LLMs and only a peculiar side-effect. I combine it with google searches (though OpenWebUI and SearXNG) and it works amazingly well then.
[1] https://www.nytimes.com/2024/09/27/technology/openai-chatgpt...
Maybe we can have a service that's an LLM with shared/cached responses as well as training on its own questions/answers for the easy stuff. Currently my usage is largely as a better search engine as google has gotten so much worse than it ever was, though I imagine I'll find custom things it's useful for as well.
At best, theres a slow march to incremental improvements that look exactly like how human culture developed knowledge.
And all the downsides will remain, the same way people, despite hundreds of good sourxes of info still prefer garbage.
> If he sincerely thought AGI was on the horizon he wouldn't be eyeing the exit
If such a thing could exist and was right around the corner, why would you need a company for it? Couldn't the AGI manage itself better than you could? Job's done, time to get a different hobby.
AGI doesn’t mean smarter than the best humans.
I'm a bit tired of the hype surrounding LLMs, but all the same for very mundane and humbler tasks that require some intelligence modern LLMs manage to surprise me on a daily basis.
But it rarely accomplishes more than what a small collection of humans with some level of expertise can achieve, when asked.
Surely, the LLM models we have today are astounding by any measure, relative to just a few years ago.
But pronouncements of how this will lead to utopia, without introducing a major revision of economic arrangements, are completely, and surely intentionally/conveniently (Sam isn't an idiot) misleading.
Is OpenAI creating a class of stock so everyone can share in their gains? If not, then AGI owned by OpenAI will make OpenAI shareholders rich, very much to the degree its AGI eliminates human jobs for itself and other corporations.
How does that, as an economic situation, result in the general population being able to do anything beyond be a customer, assuming they can still make money in some way not taken over by AGI?
Utopia needs an actual plan. Not a concept of a plan.
The latter just keeps people snowed and calm before an historic level rug pull.
General intelligence would be like an impulsive 12 year old boy who could see 6 spatial dimensions and regarded us as cartoons for only sticking to 3.
I've seen some use "super" (as in superhuman) intelligence lately to describe what you're getting at.
But if one has {a, b, c} and the other has {b, c, d} neither is more or less intelligent than the other, they just have different capabilities. "Super" is a bit too one-dimensional for the job.
The Lem story "Golem XIV" concerns a machine which claims it possesses categorically superior intelligence, and further that another machine humans have built (which runs but seems unwilling to communicate with them at all) is even more intelligent still.
Golem tries to explain using analogies, but it's apparent that it finds communicating with humans frustrating, the way it might be frustrating to try to explain things to a golden retriever. Lem wrote elsewhere that the single commonality between Golem's intelligence and our own is curiosity, unlike Annie, Golem is curious about the humans which is why it's bothering to communicate with them.
Humans (of course) plot to destroy both machines. Annie eliminates the conspirators, betraying a hitherto un-imagined capability to act at great distance, and the story remarks that it seems she does so the way a human would swat a buzzing insect. She doesn't fear the humans, but they're annoying so she destroyed them without a thought.
What could you do at 12, with half of these advantages? Choose any of them, then give yourself infinite time to use them.
Many believe that AGI will happen in robots, and not in online services, simply because interacting with the environment might be a prerequisite for developing consciousness.
You mentioned boredom, which is interesting, as boredom may also be a trait of intelligence. An interesting question is if it will want to live at all. Humans have all these pleasure sensors and programming for staying alive and reproducing. The unburdened AGI in your description might not have good reasons to live. Marvin, the depressed robot, might become real.
We can't even define what consciousness is yet, let alone whats required to develop it.
Take a look at this paper: https://osf.io/preprints/osf/mtgn7
Technically no, but practically...
12 year old limitations are: A. gets tired, needs sleep B. I/O limited by muscles
Probably there are more, but if 12 year old could talk directly to electric circuits and would not need sleep or even a break, then that 12 year old would be leaps and bounds above the best human in his field of interest.
(Well motivation to finish the task is needed though)
Why would you sell that?
For these models today, if we measure the amount of energy expended for training and inference how do humans compare?
My best guess is 120,000 times more for training GPT-4 (based on claim it cost $63 million and that was all electricity at $0.15/kWh and looking only at the human brain and not the whole body).
But also, 4o mini would then be a killowatt hour for a million tokens at inference time, by the same assumptions that's 50 hours or just over one working week of brain energy consumption. A million tokens over 50 hours is 5.5 tokens per second, which sounds about what I expect a human brain to do, but caveat that with me not being a cognitive scientist and what we think we're thinking isn't necessarily what we're actually thinking.
Humans consume about 100W average power (2000 kcal to watt hours/24 hours). So 8 billion people consume ~800 GW. Call it 1 TW. Average world electric power generation is 28000 TWh / (24*365 hours) ~3 TW.
Well, you still have to have the baby, and raise it a little. And wouldn't you still want to be known as the parent of such a bright kid as AGI? Leaving early seems to be cutting down on his legacy, if a legacy was coming.
For starters, we still need the AI (LLMs for now) to be more efficient, i.e. not require a datacenter to train and deploy. Yes, I know there are tiny models you can run on your home pc, but that's comparing a bycicle to a jet.
Second, for an AGI it meaningfully improve itself, it has to be smarter than not just any one person, but the sum total of all people it took to invent it. Until then no single AI can replace our human tech sphere of activity.
As long as there are limits to how smart an AI can get, there are places where humans can contribute economically. If there is ever to be a singularity, it's going to be a slow one, and large human AI vompanies will be part of the process for many decades still.
https://en.wikipedia.org/wiki/Gartner_hype_cycle
It just keeps happening over and over. I'd say we are at "Negative press begins".
(In the Gell-Mann amnesia sense, make sure you take careful note of who was going "OAI has AGI internally!!!" and other such nonsense so you can not pay them any mind in the future)
[0] https://www.bloomberg.com/news/articles/2024-05-17/openai-di...
Eliminating Ilya's team was part of the post-coup restructuring to consolidate Altman's power, and every tribute to "safety" paid since then is spin.
Definitions of dissolve:
to cause to disperse or disappear
to separate into component parts
to become dissipated (see DISSIPATE sense 1) or decompose
BREAK UP, DISPERSE
Those seem like pretty accurate descriptions of what happened. Yes, dissolve can also mean something stronger, so perhaps it is fair to call the statement ambiguous. But it isn’t incorrect.
For many people, sadly, one can never be rich enough. My point is, planning for both short term exit, and long term gains, is essentially the same in this particular situation. What a boon! Nice problem to have!
Yes, he has made lots of investments over the years as head of YC but not every investment was successful. This was discussed on BBC's podcast 'Good Billionaire Bad Billionaire' recently.
I hear "He will find a way to not only survive but to thrive in difficult situations"
But I also hear "He will eat people to get ahead".
If the end goal is monetization of ChatGPT with ads, it will be enshittified to the same degree as Google searches. If you get to that, what is the benefit of using ChatGPT if it just gives you the same ads and bullshit as Google?
Also, don't forget the recent Apple partnership [1], a very strong signal of their strategic positioning. Aligning with Apple reinforces their credibility and opens up even more opportunities for innovation and expansion, beyond just monetizing through ads. I just searched through this thread, and it seems the Apple partnership isn't being recognized as a significant achievement under Sam Altman's tenure as CEO, which is surprising given its importance.
High-intelligence AGI is the last human invention — the holy grail of technology. Nothing could be more ambitious, and if we know anything about Altman, it is that his ambition has no ceiling.
Having said all of that, OpenAI appears to be all in on brute-force AGI and swallowing the bitter lesson that vast and efficient compute is all you need. But they’ve overlooking a massive dataset that all known biological intelligences rely upon: qualia. By definition, qualia exist only within conscious minds. Until we train models on qualia, we’ll be stuck with LLMs that are philosophical zombies — incapable of understanding our world — a world that consists only of qualia.
Building software capable of utilizing qualia requires us to put aside the hard problem of consciousness in favor of mechanical/deterministic theories of consciousness like Attention-Schema Theory (AST). Sure, we don’t understand qualia. We might never understand. But that doesn’t mean we can’t replicate.
I’m pretty sure it means exactly that. Without actually understanding subjective experience, there’s a fundamental doubt akin to the Chinese room. Sweeping that under the carpet and declaring victory doesn’t in fact victory make.
Ironically, we understand consciousness perfectly. It is literally the only thing we know — conscious experience. We just don’t know, yet, how to replicate it outside of biological reproduction.
That’s the hard part!
> It is literally the only thing we know — conscious experience. We just don’t know, yet, how to replicate it outside of biological reproduction.
This conflates subjective experience as necessary (an uncontroversial observation) with actually understand what subjective experience is.
Or put another way: we all know what it’s like to breathe, but this doesn’t imply knowledge of the pulmonary system.
I think a better analogy would be vision. Even with a full understanding of the eye and visual cortex, one can only truly understand vision by experiencing sight. If we had to reconstruct sight from scratch, it would be more important to experience sight than to understand the neural structure of sight. It gives us something to aim for.
We basically did that with language and LLMs. Transformers aren’t based on neural structures for language processing. But they do build upon the intuition that the meaning of a sentence consists of the meaning that each word in a sentence has in relation to every other word in a sentence — the attention mechanism. We used our experience of language to construct an architecture.
I think the same is true of qualia and consciousness. We don’t need to know how the hardware works. We just need to know how the software works, and then we can build whatever hardware is necessary to run it. Luckily there’s theories of consciousness out there we can try out, with AST being the best fit I’ve seen so far.
Citation?
...or are you just assuming that AGI will be able to solve all of our problems, appropos of nothing but Sam Altman's word? I haven't seen a single credible study suggest that AGI is anything more than a marketing term for vaporware.
" High-intelligence AGI is the last human invention" What? I could certainly see all kinds of entertaining arguments for this, but to write it so matter of fact was cringe inducing.
What? Would you mind explaining this?
I don’t think this is a controversial take. Many people take issue with the premise that artificial intelligence will surpass human intelligence. I’m just pointing out the logical conclusion of that scenario.
Likewise (silicon based) AGI may be so costly that it exists only for a few years before it's unsustainable no matter the demand for it. Much like Bitcoin, at least in its original incarnation.
I really don't see any reason to believe "AGI" won't just be retreading the same thoughts humans have already had and documented. There is simply no evidence suggesting it will produce truly novel thought.
Maybe not, since Altman pretty much said they no longer want to think it terms of "how close to AGI?". Iirc, he said they're moving away from that and instead want to move towards describing the process as hitting new specific capabilities incrementally.
I still don't get the safety team. Yes, I understand the need for a business to moderate the content they provide and rightly so. But elevating the safety to the level of the survival of humanity over a generative model, I'm not so sure. And even for so-called preventing harmful content, how can an LLM be more dangerous than the access of the books like The Anarchist Cookbook, the pamphlets on how to conduct guerrilla warfare, the training materials of how to do terrorisms, and etc? They are easily accessible on the internet, no?
> Superintelligence will be the most impactful technology humanity has ever invented, and could help us solve many of the world’s most important problems. But the vast power of superintelligence could also be very dangerous, and could lead to the disempowerment of humanity or even human extinction.
> How do we ensure AI systems much smarter than humans follow human intent?
This is a question that naturally arises if you are pursuing something that's superhuman, and a question that's pointless if you believe you're likely to get a really nice algorithm for solving certain kinds of problems that were hard to solve before.
Getting rid of the superalignment team showed which version Altman believes is likely.
It won't do Sam Altman and friends any good if they are the richest corpses after an unaligned AI goes rogue.
So it would be in their were egoistical self interest to make sure it doesn't.
Altman's actions are even more consistent with total confidence & dedication to a vision where OpenAI is the 1st and faraway leader in the production of the most valuable 'technology' ever. Plus, a desire to retain more personal control over that outcome – & not just conventional wealth! – than was typical with prior breakthroughs.
I'm a software engineer comfortably drawing a decent-but-not-FAANG paycheck at an established company with every intention of taking the slow-and-steady path to retirement. I'm not projecting, I promise.
> to a vision where OpenAI is the 1st and faraway leader in the production of the most valuable 'technology' ever
Except that OpenAI isn't a faraway leader. Their big news this week was them finally making an effort to catch up to Anthropic's Artifacts. Their best models do only marginally better in the LLM Arena than Claude, Gemini, and even the freely-released Llama 3.1 405B!
Part of why I believe Altman is looking to cash out is that I think he's smart enough to recognize that he has no moat and a very small lead. His efforts to get governments to pull up the ladder have largely failed, so the next logical choice is to exit at peak valuation rather than waiting for investors to recognize that OpenAI is increasingly just one in a crowd.
I don't see how it could happen at current valuation. Who would want to pay $150 bil. for an OpenAI, where Sam is no longer part of it?
It can be an "exit" in that it makes Sam extremely wealthy while he still hangs on as CEO.
Sure, that's one option. But it's pretty rare for a founder of a $10+ billion startup to immediately liquidate their shares right after an IPO.
It's also a risky bet. If OpenAI goes public, they'll have a hard time justifying the same or a higher valuation than they got Microsoft and Nvidia to buy in at.
You think "smart" implies a low chance of world-epochal AI breakthroughs that would ruin your "slow-and-steady path to retirement". You think OpenAI has "no moat". You think it's wise to satisfice – taking a "decent" paycheck – rather than shoot-for-the-moon (and beyond it, control of the 'lightcone').
That's closer to the "short-term exit" mindset – get a good-enough payout, in mundane dollars, while you can – that you're alleging in Altman than what I'm seeing in his statements & behaviors. Yes, someone with your beliefs would in fact be trying to pump OpenAI's short-term hype for a timely exit and an even-more-cushy retirement.
But Altman's already a billionaire with any creature comfort he'd want, for the rest of his life, plus at least one "doomsday bunker" (per a 2016 'New Yorker' profile).
For at least 10 years, he's been consistently expressing a desire to advance, and control, breakthroughs in AI, AGI, and ASI that could – in a true believers' worldview – wipe away as obsolete all his prior billions – as well as the "slow-and-steady path[s] to retirement" pursued by others.
He knows as an insider and contributor OpenAI's actual lead over others – who have tended to approximately-reach OpenAI's capabilities a year or more later.
That you rate as OpenAI's "big news this week" their "Artifacts" end-user feature, rather than everything else announced & rolled out, plus OpenAI's gigantic leads in funds raised and revenue run-rate, suggests a particular world-view.
I would suggest that you are over-focused on the kind of small-stakes products/markets/ambitions that are highly-legible to a "software engineer comfortably drawing a decent-but-not-FAANG paycheck at an established company" – but such considerations are barely afterthoughts in the game of power & legacy that Altman is now playing.
An intuition pump is to think of a religious zealot that has a drastically different worldview and aims and drives from you that seem completely nonsensical (e.g. perhaps not caring about wealth accumulation at all - even for stability - and only about perpetuating God's glory).
Sam is no religious zealot, but the difference in the life game he is playing and the life game most of us are playing is as vast as that between us and the zealot.
Where did that assertion come from? Has anyone come close to replicating either of these yet (other than possibly Google, who hasn't fully released their thing yet either), let alone "quickly"? I wouldn't be surprised if it's these "sideways" architectural changes actually give OpenAI a deeper moat than just working on larger models.
They both produce garbage code in this solution. Claude versions is just 20% less garbage, but still useless. The code mixes those two, even if i specify i want the python3 version or directly specifying a version.
a) "the technology is overhyped", based on some meaningless subjective criteria, if you think a technology is overhyped, don't invest your money or time in it. No one's forcing you.
b) "child abuse problems are more important", with a link to an article that clearly specifies that the child abuse problems have nothing to do with OpenAI.
c) "it uses too much energy and water". OpenAI is paying fair market price for that energy and what's more the infrastructure companies are using those profits to start making massive investments in alternative energy [1]. So if everything about this AI boom fails what we'll be left with is a massive amount of abundant renewable energy (the horror!)
Probably the laziest conjecture I have endured from The Atlantic.
[1]: https://www.cbc.ca/news/canada/calgary/artificial-intelligen...
Except that someone has to pay for it. AI companies are only willing to pay for power purchase agreements, not capital expenses. Same with the $7T of chip fab. Invest your money in huge capital expenditures and our investors will pay you for it on an annual basis until they get tired of losing money.
I absolutely support AI companies signing as many PPAs for low carbon energy even if they implode in the future. The PV panels, wind turbines, and potentially stationary storage will already be deployed at that point.
https://betterbuildingssolutioncenter.energy.gov/financing-n...
Illinois provides subsidies to their nuclear reactors because they are low carbon [1], the federal government is subsidizing Diablo Canyon in California [2] and Palisades in Michigan [3]. Every coal fired generator in the US is more expensive to run than to replace with low carbon renewables except the one in Dry Fork, WY, [4] so subsidies are just arguments, not real economic signals.
TLDR We're kicking the can until we can get more clean energy online, and that ramp rate continues to accelerate [5].
[1] https://www.cnbc.com/2021/11/20/illinois-nuclear-power-subsi... ("Why Illinois paid $694 million to keep nuclear plants open")
[2] https://www.reuters.com/world/us/us-finalizes-11-billion-cre... ("US finalizes $1.1 billion in credits for California nuclear plant")
[3] https://news.ycombinator.com/item?id=41696884 (HN: MI nuclear plant finalizes fed loan for first reactor restart in US history")
[4] https://news.ycombinator.com/item?id=37601970 (citations)
[5] https://news.ycombinator.com/item?id=41602799 (citations)
If a counterparty is just a bit too rational, move on, lots of other potentials. Markets are sentiment driven. When there is potentially irrational exuberance, leverage it.
At best, you're forcing old generation capacity that would have been retired to stay online. At worst you're forcing the government to take loans to invest in new capacity you may not be around to pay for in a few years, leaving the public finances holding the ball.
Slight nit: a board can't start a coup because a coup is an illegitimate attempt to take power and the board's main job is to oversee the CEO and replace them if necessary. That's an expected exercise of power.
The coup was when the CEO reversed the board's decision to oust him and then ousted them.
* They burned up the hype for GPT-5 on 4o and o1, which are great step changes but nothing the competition can't quickly replicate.
* They dissolved the safety team.
* They switched to for profit and are poised to give Altman equity.
* All while hyping AGI more than ever.
All of this suggests to me that Altman is in short-term exit preparation mode, not planning for AGI or even GPT-5. If he had another next generation model on the way he wouldn't have let the media call his "discount GPT-4" and "tree of thought" models GPT-5. If he sincerely thought AGI was on the horizon he wouldn't be eyeing the exit, and he likely wouldn't have gotten rid of the superalignment team. His actions are best explained as those of a startup CEO who sees the hype cycle he's been riding coming to an end and is looking to exit before we hit the trough of disillusionment.
None of this is to say that AI hasn't already changed a lot about the world we live in and won't continue to change things more. We will eventually hit the slope of enlightenment, but my bet is that Altman will have exited by then.