As Ed Zitron posits, the AI companies themselves are not too important to let them fail, but the banks and insurance companies and mutual funds and and and that have invested and will lose money when it all goes boom?
Luckily, trunp already paid out to all the ultra rich, so all that's left is the giantist debt the world has ever seen, tripled, with which to bail them out.
Watson was AI. That was the point. An often quoted summary: "IBM promoted Watson as a revolutionary tool capable of transforming oncology, creating expectations that far exceeded the system’s actual performance". Sound familiar?
For the investors who just committed billions of dollars to OpenAI and see that this is what they chose to spend me on must be eye-watering. What kind of reality distortion field must they be living in?
These grifters all think alike. Suleyman is moving the goal posts like Bone Spurs is for his Iran war; his original goal posts were "unconditional surrender." Take note, all you corrupt grifters, how well did that work out?
Recently Altman said AGI was a “useless” term and their benchmark was now (1) having more data inside data centers than outside data centers, and (2) that no leader, president, CEO would be able to do their job without the help of AI. Totally nonsensical and really showed Altman’s grift.
Redefining difficult problems into simple ones is an ongoing story in AI. Take a hard problem and redefine it as a simple one, solve the simple one, and declare victory on the hard one. The whole "language models are intelligent" position is just an extension of this scheme. Talk like you understand, think, and reason, and most people won't be able to or won't bother to tell that it is just intelligence theater. Headlines read: We have conquered AGI! Use these models to control agents who have access to your credit cards and hope for the best.
Does OpenAI have an actual plan to achieve profitability? Does anyone there actually recognize that they have built language models, but market them as intelligence models? Does anyone there actually have an idea for how scale could possibly lead to AGI, or even just to intelligence?
The first AI winter was during the mid-70s followed by another that spanned the late-80s and early-90s and now another appears to be on the way but this one will be different because the AI boom coincides with the rise of extreme levels of wealth and income inequality. The top 20% of consumers account for over 50% of consumption which has been hidden by AI investment in data centers and GPUs. We could be in a recession now if not for that. There is a chance that this could be a long winter that takes the economy along with it. There will be a lot of wealth destruction when investment in AI collapses because they overpromised and underdelivered...again.
"There will be a lot of wealth destruction when investment in AI collapses because they overpromised and underdelivered...again." Which is *not* the fault of technique per se, but the immense greed, blindness and outright stupidity which came *after* the engineers left. As Ed Zitron wrote: A prime example of the era of the "business idiots" at the helm on all sides of the businesses concerned.
"There were never any real engineers or scientists working on LLMs to begin with" You are destroying my last illusions, stop it ;-)
I just hope there were some technical expertise and sense of responsibility around at least at the beginning of the LLM area. The official turning point might have been the firing and then re-hiring of the talented Mr. Altmann.
Just so I don’t destroy all your hopes: The actual inventor of the basic technique behind LLM’s (next word prediction) — Claude Shannon (father of information theory) actually WAS a real scientist and a top notch one at that
But he did his work long long before the current AI folks adopted his idea (with no attribution?) and implemented it on a large scale
Good old Shannon. During my BA in Elektrotechnik, he played a prominent role in HF analog/digital design. I worked a few years on HF synthesizers, with A/D converters sampling the down-converted input signals.
The narrative control story and the revenue story are the same story. OpenAI buys a media outlet the same week the DoD gave Anthropic an ultimatum, strip safety guardrails for autonomous targeting or lose the contract. The "show me the money" phase turns out to require "show me the weapons."
Every disruptive technology faces the same two outcomes. It stays decentralized and outside the existing power structure, or it gets absorbed by it. Both change in the process, but the existing power maintains control. The AI labs that started as safety-focused research organizations are now defense contractors negotiating guardrail removal. The media outlet that covered the industry is now owned by the industry. Redefining superintelligence is just what absorption looks like from the inside.
In the spirit of Suleyman, I redefine super-human athleticism not as being able to outrun or outjump every human being, but as being able to walk to the fridge and back to retrieve a can of beer.
Suleyman redefining superintelligence as “delivering product value for enterprises” is the most revealing quote of the year. It tells you everything about where the incentives actually are — not moonshot AGI, but monetizable capability improvements. The real story is the gap between what gets funded and what gets hyped.
When I use a word like “superintelligence”,' Humpt-AI Dumpt-AI said in rather a scornful tone, 'it means just what I choose it to mean — neither more nor less.''
“Humpt-AI Dumpt-AI”: often commenting under the pseudosciencenym “Mustafa Suleyman”, “Sam Altman”, “Dario Amodei” Et Al
Thanks for the article. I had a good laugh at "Prominent venture capitalists like Marc Andreessen and Peter Thiel have claimed the mainstream news media does not give the tech industry a fair shake." The poor things, every news outlet is constantly scrutinizing every word they say.... /SARC! This in the age of the "CEO said a thing"-journalism.
On the "however" side of things, however, I don't think that ALL journalism/journalists can rightly be painted with that brush, at least not in every case. We ARE in an age of extremes which (if I know extremes) cannot last forever, being constantly pressured by the real existence of the longer-arch, and for which even the stout-hearted journalist has trouble writing for without fudging lines that only exist in the temporariness of the extremist mind and especially ones that deny the basics of their own humanity.
The article, however, at least raises some essential questions. (Though you are probably thinking a simple "you're welcome" would do. :o)
Thinking about this statement and focusing on "real, constructive conversation", to me that means keeping Altmann, Amodei, Suleyman and all the others off air.
We're one news cycle away from AGI being defined as figuring out how to reopen the Strait of Hormuz.
I think that would be superduperintelligence.
"No one knew opening the Strait was so complicated." -- Numbnuts Nero
Any bank that has loaned a single cent to any of these AI companies in the last year deserves to lose it all and then some.
“Loaned…” Would you be surprised if they would go bankrupt and then the state would declare that they are too important to let them fail?:)
That's exactly the problem.
As Ed Zitron posits, the AI companies themselves are not too important to let them fail, but the banks and insurance companies and mutual funds and and and that have invested and will lose money when it all goes boom?
Luckily, trunp already paid out to all the ultra rich, so all that's left is the giantist debt the world has ever seen, tripled, with which to bail them out.
Yes, but it will be you and me who will ultimately cover these losses.
Hahahahah... AI will be the next Watson!
And look at what happened to IBM.
Watson was AI. That was the point. An often quoted summary: "IBM promoted Watson as a revolutionary tool capable of transforming oncology, creating expectations that far exceeded the system’s actual performance". Sound familiar?
For the investors who just committed billions of dollars to OpenAI and see that this is what they chose to spend me on must be eye-watering. What kind of reality distortion field must they be living in?
These grifters all think alike. Suleyman is moving the goal posts like Bone Spurs is for his Iran war; his original goal posts were "unconditional surrender." Take note, all you corrupt grifters, how well did that work out?
Hey, Twitter started with podcasts so maybe that’s a smart move! 🤣
Recently Altman said AGI was a “useless” term and their benchmark was now (1) having more data inside data centers than outside data centers, and (2) that no leader, president, CEO would be able to do their job without the help of AI. Totally nonsensical and really showed Altman’s grift.
Especially as some people* (In german: https://www.heise.de/hintergrund/c-t-Story-Das-neue-Modell-11060412.html) entertain the thought that the C-level personnel is the prime target to be replaced by *today's* LLMs.
*This includes me ;-)
Redefining difficult problems into simple ones is an ongoing story in AI. Take a hard problem and redefine it as a simple one, solve the simple one, and declare victory on the hard one. The whole "language models are intelligent" position is just an extension of this scheme. Talk like you understand, think, and reason, and most people won't be able to or won't bother to tell that it is just intelligence theater. Headlines read: We have conquered AGI! Use these models to control agents who have access to your credit cards and hope for the best.
For a different take on intelligence, see: https://herbertroitblat.substack.com/p/blind-squirrel-seeks-a-theory-of. I argue there that intelligence is a theoretical entity, not a statistical summary of successful tasks.
Does OpenAI have an actual plan to achieve profitability? Does anyone there actually recognize that they have built language models, but market them as intelligence models? Does anyone there actually have an idea for how scale could possibly lead to AGI, or even just to intelligence?
“Clever AI”
Clever Hans has fooled the crowd
Counting on his hoof is loud
Everyone has missed the cues
From the owner, subtle “Phew!”’s
“Out on a LIM”
LLM is what they built
LIM is what they billed
LLM may take them far
But on a limb is where they are
LIM : Large Intelligence Model
I totally agree with your take and immensely enjoyed reading your blind squirrel paper -- it's excellent!
If you can’t convince the people alter the narrative on them. Welcome to America land of psychological operations.
True. As Gary indicated, this is exactly what OpenAI is hoping for with its somewhat desperate purchase of TBPN. It won't work.
The first AI winter was during the mid-70s followed by another that spanned the late-80s and early-90s and now another appears to be on the way but this one will be different because the AI boom coincides with the rise of extreme levels of wealth and income inequality. The top 20% of consumers account for over 50% of consumption which has been hidden by AI investment in data centers and GPUs. We could be in a recession now if not for that. There is a chance that this could be a long winter that takes the economy along with it. There will be a lot of wealth destruction when investment in AI collapses because they overpromised and underdelivered...again.
"There will be a lot of wealth destruction when investment in AI collapses because they overpromised and underdelivered...again." Which is *not* the fault of technique per se, but the immense greed, blindness and outright stupidity which came *after* the engineers left. As Ed Zitron wrote: A prime example of the era of the "business idiots" at the helm on all sides of the businesses concerned.
Unfortunately, it’s not that the engineers and scientists have left the AI companies.
There were never any real engineers or scientists working on LLMs to begin with.
Engineering and science are fundamentally about detailed understanding. And about transparency/openness.
Beyond the basics (next token prediction), the folks working on LLMs lack a real detailed understanding of how LLMs work.
And they are the exact opposite of transparent/open. They hide everything (including all the stolen IP, not incidentally)
"There were never any real engineers or scientists working on LLMs to begin with" You are destroying my last illusions, stop it ;-)
I just hope there were some technical expertise and sense of responsibility around at least at the beginning of the LLM area. The official turning point might have been the firing and then re-hiring of the talented Mr. Altmann.
Just so I don’t destroy all your hopes: The actual inventor of the basic technique behind LLM’s (next word prediction) — Claude Shannon (father of information theory) actually WAS a real scientist and a top notch one at that
But he did his work long long before the current AI folks adopted his idea (with no attribution?) and implemented it on a large scale
Stole Shannon’s idea might be more apt.
The current AI crowd never give anyone else credit for anything.
They just shamelessly steal everyone’s creations
Good old Shannon. During my BA in Elektrotechnik, he played a prominent role in HF analog/digital design. I worked a few years on HF synthesizers, with A/D converters sampling the down-converted input signals.
Altman is in a long line of “distinguished” Stanford dropouts
Elon Musk is also in that line as is Elizabeth Holmes.
I said “distinguished” , but not what by.
The narrative control story and the revenue story are the same story. OpenAI buys a media outlet the same week the DoD gave Anthropic an ultimatum, strip safety guardrails for autonomous targeting or lose the contract. The "show me the money" phase turns out to require "show me the weapons."
Every disruptive technology faces the same two outcomes. It stays decentralized and outside the existing power structure, or it gets absorbed by it. Both change in the process, but the existing power maintains control. The AI labs that started as safety-focused research organizations are now defense contractors negotiating guardrail removal. The media outlet that covered the industry is now owned by the industry. Redefining superintelligence is just what absorption looks like from the inside.
In the spirit of Suleyman, I redefine super-human athleticism not as being able to outrun or outjump every human being, but as being able to walk to the fridge and back to retrieve a can of beer.
Suleyman redefining superintelligence as “delivering product value for enterprises” is the most revealing quote of the year. It tells you everything about where the incentives actually are — not moonshot AGI, but monetizable capability improvements. The real story is the gap between what gets funded and what gets hyped.
When I use a word like “superintelligence”,' Humpt-AI Dumpt-AI said in rather a scornful tone, 'it means just what I choose it to mean — neither more nor less.''
“Humpt-AI Dumpt-AI”: often commenting under the pseudosciencenym “Mustafa Suleyman”, “Sam Altman”, “Dario Amodei” Et Al
“AIice In AI-land”
Down the rabbit hole we go
Where it stops, we’ll never know
Maddest Hatter ain’t got naught
On the blathering AI lot
Definitions are great. By my definition, I'm awesome! And Claude agrees! Everything's fine! :-)
Gift Article 04-03-26: Here's the NY Times story this morning about AI-PR. The comments section is interesting:
https://www.nytimes.com/2026/04/02/technology/openai-buys-tbpn.html?unlocked_article_code=1.YFA.luWQ.FHyeugJOQVgQ&smid=url-share
Thanks for the article. I had a good laugh at "Prominent venture capitalists like Marc Andreessen and Peter Thiel have claimed the mainstream news media does not give the tech industry a fair shake." The poor things, every news outlet is constantly scrutinizing every word they say.... /SARC! This in the age of the "CEO said a thing"-journalism.
To Thomas Schmid: I hear that.
On the "however" side of things, however, I don't think that ALL journalism/journalists can rightly be painted with that brush, at least not in every case. We ARE in an age of extremes which (if I know extremes) cannot last forever, being constantly pressured by the real existence of the longer-arch, and for which even the stout-hearted journalist has trouble writing for without fudging lines that only exist in the temporariness of the extremist mind and especially ones that deny the basics of their own humanity.
The article, however, at least raises some essential questions. (Though you are probably thinking a simple "you're welcome" would do. :o)
"The article, however, at least raises some essential questions." Oh, it does, just too timid ones, IMHO. But maybe that's just me ....
Here's the above article's heading:
OpenAI Buys Streaming Show ‘TBPN,’ Aiming to Change Narrative on A.I.
OpenAI said the deal would help it “create a space for a real, constructive conversation about the changes A.I. creates.”
Thinking about this statement and focusing on "real, constructive conversation", to me that means keeping Altmann, Amodei, Suleyman and all the others off air.
That would be a breath of fresh AIr