Tags:

Share:
Facebook

How to save your job if you're a billionaire who fails to deliver
August 19, 2025 10:43 AM Subscribe

His models don't work, his claims of Artificial Intelligence fall far behind in reality. But Sam Altman has an ace up his sleeve. In a cunningly deft maneuver, he has claimed that it's "not useful" to talk about AGI since it means different things to different people.

As far as I remember, AGI had only one popular definition - that it would be smarter than us, take our jobs and strip us of purpose and meaning in our lives.

For a guy who trained his AI on all of our content without consent, did some crooked mangling to turn a non-profit into a for-profit company, and finally could not even deliver on his promises, Sam Altman hopes to retain his job with his beautiful reframing as messiah whose mission is "grossly misunderstood" by us "mere mortals".

posted by mahadevan (59 comments total) 13 users marked this as a favorite

It would be possible to have an interesting discussion about this question but not with this post framing.
posted by Nelson at 10:51 AM on August 19 [31 favorites]

Pace Nick Cave, I do not believe in an interventionist God. But if I did, I would petition for species-level protection from charismatic charlatans. I truly cannot think of a more dangerous thing in this world.

Considering that, I fully expect Altman to keep his position. He's got these peoples' number coming and going, and they're too invested in him to pitch him over the side. It would mean admitting they've been wrong this whole time, and I have trouble thinking of any time in human history when a group of rich and powerful people did a good job of that. Not saying it never happens, but it's pretty rare. Doubtful it happens here.
posted by Smedly, Butlerian jihadi at 10:53 AM on August 19 [13 favorites]

From “the spectral evidence for AI”:

Tech is a gated community. They are part of the world, but imagine themselves different, above, and better. All gated communities operate in the same way, whether real or spiritual. All people within the community do, say, believe, eat, look, and believe essentially the same things, but this is treated as extreme individuality that, again, separates and makes you better than everyone else. Whether explicitly religious or nominally secular, all gated communities rely on shared values and beliefs that essentially turn them into cults.

The bubble of AI is collapsing, but the tech guys are still a dominant ideological force. Just like how failed politicians get sinecures, some of the AI guys are going to be slotted into the tech ranks as “visionaries”. Clearly, Sam Altman would like that to include him.
posted by The River Ivel at 11:01 AM on August 19 [9 favorites]

Remember how Elon used to claim that self-driving cars would exist next year? and he said that every year for like ten years? This reminds me of that.
posted by nushustu at 11:02 AM on August 19 [14 favorites]

Sam Altman, the guy who thinks we should put a Dyson Sphere around the solar system to house all the data centers needed to get to AGI?

He's a huckster. The difference between Sam Altman and a Snake Oil Salesmen is scale, not content.
posted by Frayed Knot at 11:06 AM on August 19 [27 favorites]

AGI is just "A Gigadollar Investment" and nothing more. We don't have any collective concept of how to qualitatively or quantitatively measure whatever sort of digital omnipotence these guys are blathering about. AGI is simply whatever the richest person says it is, and Altman's at least cogent enough to diminish an already-fallacious finish line for AI fundraising so when Elon declares Grok to have achieved AGI because it's not woke anymore, there's already an established narrative saying it's a meaningless term anyway.
posted by phong3d at 11:24 AM on August 19 [3 favorites]

I don’t understand how we could be on the verge of awakening sentience when websites cannot even tell if I want to accept all cookies.
posted by johngoren at 11:26 AM on August 19 [41 favorites]

In a way I'm glad MIRI exists. I think it's important that there's a cult with buy-in from industry leaders that for over a decade was making prophecies that were so far outside the realm of possibility that their continual failure can be used to disqualify the entire class of the new captains of indistry from power over others. We must never forget that they are also blood and flesh.
posted by jy4m at 11:35 AM on August 19 [6 favorites]

As far as I remember, AGI had only one popular definition - that it would be smarter than us, take our jobs and strip us of purpose and meaning in our lives.

Don’t forget also torturing simulations of us for all eternity.
posted by star gentle uterus at 11:44 AM on August 19 [10 favorites]

Recently his pitches have also drifted over to "give us huge $ investment, and when we make AGI, the AGI will know how to give you huge returns."

Tech bros love reinventing trains, and buses, and even Pascal's Wager. Now they're reinventing the prosperity gospel.
posted by tclark at 12:04 PM on August 19 [23 favorites]

It’s the Roko’s basilisk of investing!
posted by los pantalones del muerte at 12:09 PM on August 19 [8 favorites]

He's saying this because he wants to get out from under Microsoft, and the condition for that happening is that OpenAI has to "achieve general AI." This is a condition that Altman presumably wrote himself.

The reason he's so desperate to become independent is that he has $10 billion from SoftBank riding on it.
posted by Just the one swan, actually at 12:15 PM on August 19 [12 favorites]

"The problem with AGI, Altman said, is that there are multiple definitions being used by different companies and individuals."

Yeah, the priests and patriarchs figured out this rhetorical technique for question dodging thousands of years ago. Like I said before: AI churches are going to be a thing. They already are a thing, if you count The Wizard of Oz as the first one.
posted by AlSweigart at 12:33 PM on August 19 [3 favorites]

AI churches are going to be a thing.

Gary wouldn't approve of that.
posted by RonButNotStupid at 12:49 PM on August 19 [3 favorites]

'AI' is a perfectly serviceable acronym. 'AGI' seems like a dickhead shibboleth.
posted by rhamphorhynchus at 1:00 PM on August 19 [1 favorite]

> In a cunningly deft maneuver, he has claimed that it's "not useful" to talk about AGI since it means different things to different people.

It's not that cunning since he's personally responsible for a lot of the conflicting definitions. This would include one in which AGI is only achieved when OpenAI becomes sufficiently profitable.
posted by at by at 1:01 PM on August 19 [6 favorites]

Chinese room wasn't just a room, Sam. It was a metaphor that you have missed.
posted by DeepSeaHaggis at 1:07 PM on August 19 [2 favorites]

Mr Altman of OpenAI,
People say you're a smart kinda guy.
But building a Dyson
Sphere? Really? Nice try, son.
And still they invest? My prompt: Why?
posted by rory at 1:57 PM on August 19 [17 favorites]

I, for one, was delighted by the framing by the OP, if you want an alternate opinion.

Instead of Roko’s basilisk it’s more like a maxwell’s demon but where it’s money not energy. This is just him shutting the door as quickly as possible to keep the dollars on his side.
posted by q*ben at 2:12 PM on August 19 [13 favorites]

To anyone with money wishing they could compete with Mister Bad Hair Day, give me carte blanche and an unlimited budget and I'll have something useful one year from when I start. They're all digging in the wrong place.
posted by ob1quixote at 2:14 PM on August 19 [2 favorites]

ChatGPT has 700 million weekly users.
posted by gwint at 2:41 PM on August 19 [3 favorites]

There's a ChatGPT user born every minute.
posted by Phssthpok at 2:51 PM on August 19 [9 favorites]

Must be more than that, at one a second it'd take 22 years.
posted by the antecedent of that pronoun at 2:57 PM on August 19 [1 favorite]

Smedly, Butlerian jihadi: "Pace Nick Cave, I do not believe in an interventionist God. But if I did, I would petition for species-level protection from charismatic charlatans. I truly cannot think of a more dangerous thing in this world."

I've often thought that if i ever got one wish from some sort of magical being, my wish would be to change human biology such that lying causes physical pain, in proportion to how big a lie it is. I don't want it to be impossible to lie; some lies are useful and important (especially the small social lies, which would only cause a moment of toothache or something)! But it should have inescapable consequences.
posted by adrienneleigh at 3:05 PM on August 19 [10 favorites]

Clearly the answer will be fusion power. Once fusion power is commercially viable and well understood we'll have the power we need to finally power AI tech that is smart enough to create improved versions of itself.

Just give it another 10-20 years and they'll have it cracked....
posted by VTX at 3:09 PM on August 19 [4 favorites]

'AI' is a perfectly serviceable acronym. 'AGI' seems like a dickhead shibboleth.

I assume it's the g of general intelligence.
posted by clew at 3:11 PM on August 19 [4 favorites]

I feel like I say some version of this in every AI thread but the thing about “AI” is that it evolved two parallel definitions: the popular one of “computers that think like people” and the technical one of “the broad subfield of computer science concerned with automated problem solving.” “AGI” offers an alternative term for the former in an attempt to avoid this ambiguity, which seems fair enough, except it still doesn’t solve the problem that when you start trying to talk about it in concrete terms nobody really agrees on what it means.
posted by atoxyl at 3:22 PM on August 19 [7 favorites]

Sam Altman, the guy who thinks we should put a Dyson Sphere around the solar system to house all the data centers needed to get to AGI?

Literally every day he says increasingly insane bullshit. I think he's having a public nervous breakdown. Frankly, I hope he is, he deserves nothing but anguish for the rest of his days.

Literally yesterday he was asking for billions to achieve AGI, so he could ask it how to achieve a profitable business model.
posted by His thoughts were red thoughts at 3:25 PM on August 19 [2 favorites]

AlSweigart: "AI churches are going to be a thing. "

Roko's Basilica
posted by chavenet at 3:46 PM on August 19 [58 favorites]

You know how some people think of LLMs as text-generating machines (e.g.: "fancy autocomplete") that merely mimic things like "knowledge" and "reasoning" through clever mimicry and adaptation of existing texts while other people genuinely believe that LLMs know and think and feel things like people do? What if the CEO of OpenAI turned out to be in the latter category? (Or maybe at least until very recently) What if Sam wasn't just putting on a huckster act to sell his snake oil? What if he was a true believer? Is it worse to be governed by a liar or a fool?
posted by mhum at 4:15 PM on August 19 [1 favorite]

At a commercial level, this is completely true. There's no market demand for general intelligence as such, be it artificial or natural. ("If you're so smart, why ain't you rich" is a fundamental...)

The market wants to pay less for the same production, or get more production for the same cost, or (if you get lucky) pay less for more production. Whatever pseudo-cognition helps for whatever kind of production, is what people want.
posted by MattD at 4:26 PM on August 19 [4 favorites]

github: 4.1 Opus Committed Deliberate Task Fraud in Production Context

previously: "[If] developers are allowed to use AI tools, they take 19% longer to complete issues—a significant slowdown that goes against developer beliefs and expert forecasts. This gap between perception and reality is striking: developers expected AI to speed them up by 24%, and even after experiencing the slowdown, they still believed AI had sped them up by 20%."
posted by jeffburdges at 4:27 PM on August 19 [4 favorites]

> Is it worse to be governed by a liar or a fool?

?Por qué no los dos? (Bua Bua Bua)

Cleo Abram had Sam Altman on Huge If True a week and a half ago, and though I almost always love her videos, I just couldn't stomach it. Altman is such a tedious blatherskite.
posted by Hot Pastrami! at 4:42 PM on August 19

OpenAI in Deal Talks That Would Value the Company at $500 Billion:

OpenAI, the maker of ChatGPT, is in talks to sell $6 billion in shares owned by its current and former employees to investors, in a deal that would value the artificial intelligence company at roughly $500 billion, according to two people with knowledge of the discussions. At $500 billion, OpenAI would become the world’s most valuable privately held company

posted by gwint at 5:19 PM on August 19

I really enjoyed reading through the comments.

I learnt more science from this thread than any of Sam Altman's intereviews lol.

I know I'm dripping sarcasm but I think I'm justified considering the humongous boatload of rubbish has been sold to us as "evolution".
posted by mahadevan at 5:24 PM on August 19 [2 favorites]

> they take 19% longer to complete issues

Yes, but the technology will keep improving and in five years maybe they'll only take 18% longer.
posted by AlSweigart at 5:38 PM on August 19 [7 favorites]

It would be possible to have an interesting discussion about this question but not with this post framing.

Cool. I'm just waiting for Sam to pull a Yahoo! and turn down a ridiculous offering, and then his company gets sold for scraps twelve months down the line.
posted by They sucked his brains out! at 5:39 PM on August 19 [2 favorites]

“The spectral evidence for AI” rocked, thanks The River Ivel. :)

If one needs better AI post framing, then I suggest the TED Talk by There I Ruined It, seemingly the most talented AI-ish musician right now.

“Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.” ― Frank Herbert, Dune
posted by jeffburdges at 6:05 PM on August 19 [3 favorites]

I have been saying for a while that the problem with people who blindly believe that the current crop of chatbots portending AI being "right around the corner" don't have enough interactions with developmentally disabled adults.

Man, I was not expecting this particular flavor of ablism in my Metafilter this evening. It's not o.k. to compare something you don't like to disabled folk. Even if you think it's particularly apt. If you're using a disability as an analogue to how bad something is at being human, you're saying disabled people aren't good at being human.

The current culture around AI sucks, but so does ablism.
posted by Gygesringtone at 6:06 PM on August 19 [18 favorites]

Do you want to know how to get an instant rejection in an interview? Things were going great until I asked the founders “what happens to your business model if OpenAI goes out of business?”

I mean, they sent the email like five minutes after the interview was over. Weird thing was, as far as I could tell, they wouldn’t actually be that impacted by OpenAI going away.

Ah well.
posted by Skrubly at 6:39 PM on August 19 [5 favorites]

Turns out that Copilot will happily let you steal information and leave no trace, if you ask it nicely.
posted by JoeZydeco at 6:50 PM on August 19 [5 favorites]

Nelson: "It would be possible to have an interesting discussion about this question but not with this post framing."

MetaTalk post regarding editorializing in this post, and in front page posts in general.
posted by brainwane at 8:23 PM on August 19 [2 favorites]

Things were going great until I asked the founders “what happens to your business model if OpenAI goes out of business?”

I chose to believe you were interviewing at SoftBank.
posted by ryanrs at 8:26 PM on August 19 [1 favorite]

I don't think it's correct to compare chatbots to humans, and I think it's actively ablelist to compare them to disabled humans specifically to highlight their dissimilarities with humans.
posted by jy4m at 9:03 PM on August 19 [7 favorites]

We cannot avoid comparing chatbots to humans entirely, since we've posts every day that do so, and this poist discusses how Sam Altman compares them to humans. Again The River Ivel's link handles that conversation better.
posted by jeffburdges at 9:32 PM on August 19

What if Sam wasn't just putting on a huckster act to sell his snake oil? What if he was a true believer?

ChatGPT seems to be having an effect on some OpenAI-adjacent people and as noted upthread there's a strong element of cult to the whole tech world these days (MIRI etc seem to be cult generators) so he might well believe the hype. Life inside OpenAI must be especially brain-breaking.
posted by BungaDunga at 9:44 PM on August 19 [1 favorite]

What if Sam wasn't just putting on a huckster act to sell his snake oil? What if he was a true believer?

He's worthy of contempt and scorn either way. To the rest of us it makes no difference, as his impact is the same.
posted by His thoughts were red thoughts at 9:58 PM on August 19 [2 favorites]

Roko's Basilica

I really want to like this, chavenet, but I also feel strongly that you've had it locked and loaded for a few years now.
posted by DeepSeaHaggis at 2:26 AM on August 20

95% of generative AI pilots at companies are failing:

The GenAI Divide: State of AI in Business 2025, a new report published by MIT’s NANDA initiative, reveals that while generative AI holds promise for enterprises, most initiatives to drive rapid revenue growth are falling flat.

Despite the rush to integrate powerful new models, about 5% of AI pilot programs achieve rapid revenue acceleration; the vast majority stall, delivering little to no measurable impact on P&L. The research—based on 150 interviews with leaders, a survey of 350 employees, and an analysis of 300 public AI deployments—paints a clear divide between success stories and stalled projects.

(Although they still try to spin this as "you're holding it wrong" by blaming the users for not being good prompt jockeys or buying the right tools or using it for the wrong thing or something)
posted by autopilot at 2:46 AM on August 20 [5 favorites]

nushustu: Remember how Elon used to claim that self-driving cars would exist next year? and he said that every year for like ten years? This reminds me of that.

Wikipedia remembers! The list of predictions for autonomous Tesla vehicles by Elon Musk reminds me of my favorite chart -- Itanium sales forecasts that get increasingly worse every year but always show exponential growth eventually (although the "actual sales" line at the bottom tells a different story).
posted by autopilot at 2:52 AM on August 20 [3 favorites]

The problem with selling the entire planet a monorail is that there isn't another planet to escape to with your carpet bags stuffed full of cash.
posted by Molesome at 3:13 AM on August 20 [3 favorites]

mahadevan: "URGENT: I don't know how else to spread this so posting here - someone can repost this as a new post to MeFI:"

Hey, just fyi MetaFilter isn't a site for "you must see this now, it's very important" aggressive type comments or posts, nor is it encouraged to ask people to post something for you. In fact, if a person is feeling that something must posted and/or as soon as possible, we advise you rethink that that line of thought before posting.

The comment was also off topic, meaning the thread is about AI/Sam Altman, but the comment was about Trump and the Smithsonian. Those two things aren't related.

So the comment has been deleted, for being a derail. Not a big deal, it's nothing personal! Just avoid posting offtopic comments in a thread, along with imploring others to make a post because you think it must be shared.
posted by Brandon Blatcher (staff) at 5:14 AM on August 20 [3 favorites]

straw: "I have been saying for a while that the problem with people who blindly believe that the current crop of chatbots portending AI being "right around the corner" don't have enough interactions with developmentally disabled adults. That willingness to tell you "yes", the infodumping on things they've read or otherwise imbibed from their environment that's almost plausible, but absolute bullshit? These are patterns that anyone who's spent time in a group home, or around programs for those with developmental delays, or just hanging out with DD relatives, picks up on real fast.
"

Hi, please be careful about you categorize and write about other people. While I don't think you met this comment to be dismissive of people that are developmentally disabled, the comment did pick up a number of flags that very understandably took it that, so heads about writing about others: be respectful and avoid using them (whoever they are) are examples to prove a point of yours. Because you may be ruining your own point be unknowingly being offensive.
posted by Brandon Blatcher (staff) at 6:05 AM on August 20 [1 favorite]

But Sam Altman has an ace up his sleeve. In a cunningly deft maneuver,

funny how one man's "cunningly deft" is another man's "shamelessly deceitful"
posted by Borborygmus at 7:35 AM on August 20 [1 favorite]

I don't think it's okay to rewrite a CNBC news article as if it were an opinion essay. This kind of behavior contributes to online misinformation, it doesn't rectify it or whatever latest evils the tech industry is up to.
posted by polymodus at 9:03 AM on August 20

> His thoughts were red thoughts: "He's worthy of contempt and scorn either way. To the rest of us it makes no difference, as his impact is the same."

True. But as a connoisseur of contempt and scorn, I like to fine-tune and calibrate my disdain.

Also, while it may not matter to us in the hoi polloi, I can't help but wonder if the people who are propping him up at Microsoft and A16Z and SoftBank (well, probably not Softbank) are asking themselves the same questions. It's one thing to back a con-man. It's another to back a con-man who got high off his own supply and ended up conning himself.
posted by mhum at 9:38 AM on August 20 [2 favorites]

Satya Nadella at least seems to have a bond with AI already:

Copilot consumes Nadella’s life outside the office as well. He likes podcasts, but instead of listening to them, he loads transcripts into the Copilot app on his iPhone so he can chat with the voice assistant about the content of an episode in the car on his commute to Redmond. At the office, he relies on Copilot to deliver summaries of messages he receives in Outlook and Teams and toggles among at least 10 custom agents from Copilot Studio. He views them as his AI chiefs of staff, delegating meeting prep, research and other tasks to the bots. “I’m an email typist,” Nadella jokes of his job, noting that Copilot is thankfully very good at triaging his messages.

I don't think there are any adults in the room; these guys are all in the same group chats and have been marinating in each other's and their AI's farts for years.
posted by BungaDunga at 10:00 AM on August 20 [3 favorites]

Mahadevan, no worries on this. I didn't realize you were new. Welcome to MeFi!
posted by storybored at 11:48 AM on August 20 [2 favorites]

Sam Altman, the guy who thinks we should put a Dyson Sphere around the solar system to house all the data centers needed to get to AGI?

This Golden Calf will become a god if we just melt down enough gold for it. We just haven't melted enough gold yet. Fortunately, many of the dudes who have hoarded most of the planet's gold are on board with this plan...
posted by straight at 10:21 PM on August 21

« Older "This was easier to get into than I thought." | The LGBTQIA+ News Post, August 19, 2025: Hi Honey... Newer »

This thread has been archived and is closed to new comments

苍井空番号迅雷链接

How to save your job if you're a billionaire who fails to deliver August 19, 2025 10:43 AM Subscribe

How to save your job if you're a billionaire who fails to deliver
August 19, 2025 10:43 AM Subscribe