Moon of Alabama Brecht quote
January 24, 2025

How The Chinese Beat Trump And OpenAI

The hype around Artificial Intelligence, the now failed U.S. attempt to monopolize it, and the recent counter from China are a lesson in how to innovate. They also show that the U.S. is losing the capability to do so.

In mid 2023, when the Artificial Intelligence hype gained headlines, I wrote:

'Artificial Intelligence' Is (Mostly) Glorified Pattern Recognition

Currently there is some hype about a family of large language models like ChatGPT. The program reads natural language input and processes it into some related natural language content output. That is not new. The first Artificial Linguistic Internet Computer Entity (Alice) was developed by Joseph Weizenbaum at MIT in the early 1960s. I had funny chats with ELIZA in the 1980s on a mainframe terminal. ChatGPT is a bit niftier and its iterative results, i.e. the 'conversations' it creates, may well astonish some people. But the hype around it is unwarranted.
...
Currently the factual correctness of the output of the best large language models is an estimated 80%. They process symbols and pattern but have no understanding of what those symbols or pattern represent. They can not solve mathematical and logical problems, not even very basic ones.

There are niche applications, like translating written languages, where AI or pattern recognition has amazing results. But one still can not trust them to get every word right. The models can be assistants but one will always have to double check their results.

Overall the correctness of current AI models is still way too low to allow them to decide any real world situation. More data or more computing power will not change that. If one wants to overcome their limitations one will need to find some fundamentally new ideas.

But the hype continued. One big AI model, ChatGPT, was provided by a non-profit organization, OpenAI. But its CEO, Sam Altman, soon smelled the big amount of dollars he potentially could make. A year after defending the the non-profit structure of OpenAI Altman effectively raided the board and took the organization private:

ChatGPT-maker OpenAI is working on a plan to restructure its core business into a for-profit benefit corporation that will no longer be controlled by its non-profit board, people familiar with the matter told Reuters, in a move that will make the company more attractive to investors.
...
Chief executive Sam Altman will also receive equity for the first time in the for-profit company, which could be worth $150 billion after the restructuring as it also tries to remove the cap on returns for investors, sources added.

The ChatGTP large language model OpenAI provided was closed source. A black-box, running in the cloud, that one could pay to chat with or use for translating, content generation or analyzing certain problems.

The training and maintaining of ChatGTP took large amounts of computing power and money. It was somewhat expensive but there was no new technology in it. The algorithms it used were well known and the training data needed to 'program' it were freely available internet content.

For all the hype about AI is is not a secret or even new technology. The barriers to entry for any competition is low.

That is the reason why Yves at Naked Capitalism, pointing to Edward Zitron, asked: “How Does OpenAI Survive?” It doesn't. Or has little chance to do so. Discussions in the U.S. never acknowledged those facts.

Politicians thought of AI as the next big thing that would further U.S. control of the world. They attempted to prevent any potential competition to the lead the U.S. thought it had in that field. Nvidea, the last leading U.S. chip maker, lost billion when it was prohibited from selling in latest AI-specialized models to China.

Two days ago Trump announced Stargate, a $500 billion AI infrastructure investment in the US:

Three top tech firms on Tuesday announced that they will create a new company, called Stargate, to grow artificial intelligence infrastructure in the United States.

OpenAI CEO Sam Altman, SoftBank CEO Masayoshi Son and Oracle Chairman Larry Ellison appeared at the White House Tuesday afternoon alongside President Donald Trump to announce the company, which Trump called the “largest AI infrastructure project in history.”

The companies will invest $100 billion in the project to start, with plans to pour up to $500 billion into Stargate in the coming years. The project is expected to create 100,000 US jobs, Trump said.

Stargate will build “the physical and virtual infrastructure to power the next generation of AI,” including data centers around the country, Trump said. Ellison said the group’s first, 1 million-square foot data project is already under construction in Texas.

On the very same day, but with much less noise, a Chinese company published another AI model:

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors.

The new DeepSeek models have better benchmarks than any other available model. They use a different combination of technics, less training data and much less computing power to achieve that. They are cheap to use and, in contrast to OpenAI, real open source.

Writes Forbes:

U.S. export controls on advanced semiconductors were intended to slow China's AI progress, but they may have inadvertently spurred innovation. Unable to rely solely on the latest hardware, companies like Hangzhou-based DeepSeek have been forced to find creative solutions to do more with less.
...
This month, DeepSeek released its R1 model, using advanced techniques such as pure reinforcement learning to create a model that's not only among the most formidable in the world, but is fully open source, making it available for anyone in the world to examine, modify, and build upon.
...
DeepSeek-R1’s performance is comparable to OpenAI's top reasoning models across a range of tasks, including mathematics, coding, and complex reasoning. For example, on the AIME 2024 mathematics benchmark, DeepSeek-R1 scored 79.8% compared to OpenAI-o1’s 79.2%. On the MATH-500 benchmark, DeepSeek-R1 achieved 97.3% versus o1’s 96.4%. In coding tasks, DeepSeek-R1 reached the 96.3rd percentile on Codeforces, while o1 reached the 96.6th percentile – although it’s important to note that benchmark results can be imperfect and should not be overinterpreted.

But what’s most remarkable is that DeepSeek was able to achieve this largely through innovation rather than relying on the latest computer chips.

Nature is likewise impressed:

A Chinese-built large language model called DeepSeek-R1 is thrilling scientists as an affordable and open rival to ‘reasoning’ models such as OpenAI’s o1.
...
“This is wild and totally unexpected,” Elvis Saravia, an AI researcher and co-founder of the UK-based AI consulting firm DAIR.AI, wrote on X.

R1 stands out for another reason. DeepSeek, the start-up in Hangzhou that built the model, has released it as ‘open-weight’, meaning that researchers can study and build on the algorithm. Published under an MIT licence, the model can be freely reused but is not considered fully open source, because its training data has not been made available.

“The openness of DeepSeek is quite remarkable,” says Mario Krenn, leader of the Artificial Scientist Lab at the Max Planck Institute for the Science of Light in Erlangen, Germany. By comparison, o1 and other models built by OpenAI in San Francisco, California, including its latest effort o3 are “essentially black boxes”, he says.

Even long term Internet investors, who have seen it all, are impressed:

Marc Andreessen 🇺🇸 @pmarca - 9:19 UTC · Jan 24, 2025

Deepseek R1 is one of the most amazing and impressive breakthroughs I’ve ever seen — and as open source, a profound gift to the world. 🤖🫡

Nature adds:

DeepSeek hasn’t released the full cost of training R1, but it is charging people using its interface around one-thirtieth of what o1 costs to run. The firm has also created mini ‘distilled’ versions of R1 to allow researchers with limited computing power to play with the model.

That does in fact work!

Brian Roemmele @BrianRoemmele - 14:34 UTC · Jan 23, 2025

Folks, I think we have done it!
If overnight tests are confirmed we have OPEN SOURCE DeepSeek R1 running at 200 tokens per second on a NON-INTERNET connected Raspberry Pi.
A full frontier AI better than “OpenAI” owned fully by you in your pocket free to use!
I will make the Pi image available as soon as all tests are complete.
You just pop it into a Raspberry Pi and you have AI!
This is just the start of the power that takes place when you TRULY Open Source an AI Model.

The latest Rasberry Pi hardware starts at $50. The software is free.

This is a death call for OpenAI:

Arnaud Bertrand @RnaudBertrand - 14:23 UTC · Jan 21, 2025

Most people probably don't realize how bad news China's Deepseek is for OpenAI.

They've come up with a model that matches and even exceeds OpenAI's latest model o1 on various benchmarks, and they're charging just 3% of the price.

It's essentially as if someone had released a mobile on par with the iPhone but was selling it for $30 instead of $1000. It's this dramatic.

What's more, they're releasing it open-source so you even have the option - which OpenAI doesn't offer - of not using their API at all and running the model for "free" yourself. ...

The backstory of DeepSeek is also amazing.

In 2007 three Chinese engineers set out to build a quant (financial speculation) fund using AI. They hired hungry people fresh from the universities. Their High-Flyer fund was somewhat successful but throughout the last years the Chinese government started to crack down on financial engineering, quant trading and speculation.

With time on their hand and unused computing power in their back room the engineers started to build the DeepSeek models. The costs were minimal. While OpenAI, Meta and Google spent billions to build their AI's the training costs for the published DeepSeek models were mere $5 to 6 million.

Henry Shi @henrythe9ths - 23:20 PM · Jan 20, 2025

7. The lesson?

Sometimes having less means innovating more. DeepSeek proves you don't need:
- Billions in funding
- Hundreds of PhDs
- A famous pedigree
Just brilliant young minds, the courage to think differently and the grit to never give up

Another lesson is that brilliant young minds should not be wasted to optimize financial speculation but to make stuff one can use.

DeepSeek demonstrates how it is impossible to use trade and technology barriers to keep technology away from competitors. They can, with decent resources, simply innovate around those.

Even billions of dollars, loud marketeers like Trump and self promoting grifters like Sam Altman can not successfully compete with a deep bench of well trained engineers.

As an author at Guancha remarks (machine translation):

In the Sino-US science and technology war, China's unique advantage comes precisely from the US ban. It can be said that our strong will to survive was forced out by Washington, and maximizing our limited resources is the secret to breaking through. In history, this kind of story is not new, that is, the weak prevail over the strong, and the small fight against the big.

The U.S. side will fall into a Vietnam-style dilemma-relying too much on its own absolute advantage, thus wasting a lot of resources and losing itself to internal consumption.

How long for the U.S. to (re-)learn that lesson?

Posted by b on January 24, 2025 at 15:46 UTC | Permalink

Comments
next page »

kudos china and the folks able to pull this off!! creativity of people, knows no bounds!!! good story b.. thanks!

Posted by: james | Jan 24 2025 15:57 utc | 1

It sounds like in China you can do radically new innovation without the permission of Silicon Valley and a dozen Federal agencies...Whodathunk it?

Posted by: pyrrhus | Jan 24 2025 16:01 utc | 2

Another win for sanctions!
Those dead-eyed globalists never learn.

Posted by: ChatNPC | Jan 24 2025 16:06 utc | 3

This event underscores the value of:

a. Lots of people doing product development. The more participants, from more discipline areas, cooperating more, gets results which exceed those of closed development environments, even when the closed environments are lavishly funded.

b. Building the Commons. Altman pursued the "build a moat around it, and extract wealth" model, even after posing as a do-gooder. Now he's going to get his business-model's clock cleaned. China contributed something terrific to the Commons. Whose actions will have greater positive impact?

This event also underscores the sheer impossibility of containing 1.2 billion people who are determined, capable, and very intelligent. There is no defense from this force, no containing it.

One U.S. strategy that might keep us in the game is to modify our culture. Reward risk-taking, resource the little people, encourage team-work, build tools and platforms and forums that facilitate collaboration.

That might work.

At one time the U.S. prided itself on being the world center of innovation. Then came financialization, whose core function was to extract wealth from products and companies that already exist. Our "innovation" centered on social media and smart phones, and delivered a culture, a population of consumers and passive observers, trading pix of their pet squirrel and today's foodie experience.

Not pix of their new invention, nor the team-photo of their virtual product development team, or their home-based development workbench, nor their new small enterprise that was built to commercialize their innovations.

We have a culture problem. We're not rewarding and resourcing the behaviors that lead to great problems being solved.

We're not engaging the minds of our 350 million people; we're wasting their intellect and their time. We celebrate football players, "thanks for your service to the Empire" soldiers, we fund the "Defense" industry whose core function is to destroy stuff and squash whole countries.

We have a culture problem: we value the wrong things, and we reward the wrong behaviors.

China values better things, and rewards the people that do those better things. Expect to see the innovation and wealth-building gap between the U.S. and China widen considerably.

Posted by: Tom Pfotzer | Jan 24 2025 16:07 utc | 4

Of course, the real problem is that AI has already been declared the new God, and any question of its infallibility is never raised. Despite the fact Open AI is shown to be highly error prone, unable to give a consistent answer to the same question submitted multiple times it is being rapidly rolled out as a means of, you guessed it, replacing human operators.

The danger is an "AI says no" dystopia against which those humans that remain can have no recourse.
(Unless we insist that the PMC of ugly bags of mostly water are the first cohort to be replaced by AI, it could hardly do a worse job).

Posted by: ChatNPC | Jan 24 2025 16:11 utc | 5

As a paid subscriber to ChatGPT

It replaces google. I ask it a question and it gives me an answer to my question. Google search gives me 4 ads followed by a list of urls that I have to parse and most of them are garbage. OpenAI is a google killer.

ChatGPT has made me into a more efficient Software developer because it gives me the correct syntax for obscure commands. 'How do I use sed to replace the 2nd to last field of blah with blah-blah'. Thank goodness I no longer have to read the entire man page for 'sed' or search
for the answer on stack overflow. I don't feel threatened by ChatGPT. It takes knowledgeable people to ask the questions.

BTW I am amazed, floored, dumbfounded, that ChaptGPT understands my word salad questions. I'm still excited about it (that's just the geek in my coming out).

China
I am tired of hearing that China simply copies us. When China builds something great, we claim that copied us by showing a comic book we published in 1957 that talked about ray guns (yes, that makes it our intellectual property :-))

Posted by: Christian J Chuba | Jan 24 2025 16:17 utc | 6

An excellent example of how Nature's secrets are open for anyone to discover and innovate:

"Just brilliant young minds, the courage to think differently and the grit to never give up."

And AI isn't the only area of science where the Outlaw US Empire is being outdistanced by that same sort of effort. This Global Times op/ed will certainly open some eyes, "Why is news of overseas talents coming to China ‘trending’?". The opening paragraph:

"Recently, the news of several outstanding talents who graduated from Peking University and Tsinghua University returning to China from abroad has attracted attention. A closer look at the websites of domestic universities reveals that headlines about the return of overseas research talent have been "trending." Many of these individuals are in the prime of their careers and have the potential to make substantial contributions. Some are chief scientists at renowned international companies, while others have secured tenured positions at world-class research institutions. Behind the "trending" lies a resonance among various talents regarding China's innovation ecosystem, as well as a significant test of the openness and innovative capabilities of this great Eastern technological power. The curtain has already risen on a new era of globalization, and China is actively participating, poised to exert an even greater siphoning effect on this wave."

The human capital involved is where the long run advantage lies, and US policy has worked negatively in that manner by shunning non-US native minds through various mechanisms, including Trump's War on Immigrants.

Posted by: karlof1 | Jan 24 2025 16:17 utc | 7

That remind me my schooldays , we had once a project to do with an 8 bits processor , I did it with a ISP GAL at a fraction of the cost. I was mocked and got 0 points on the theoretical part of the project ... then came the prototype part and my cheap board was the only one doing the job it was meant for in the whole classroom. I still scored 0 point but I couldn't have cared less : mine worked.

Posted by: Savonarole | Jan 24 2025 16:24 utc | 8

I think most of you guys never actually used the most advanced OpenAI model. o1 is actually pretty solid and consistent. Claude is also an impressive model with amazing reasoning capabilities, albeit not on this high level.

What b somehow forgets is that OpenAIs o3 is around the corner. So yes, there is now a chinese Model wich can rival o1. But wait a month and its not so interesting anymore.

For the future, I can foresee a mixed approach, which OpenAI will not really like:
use something like DeepSeek-R1 for pre-processing a certain problem or document - and then use the highest-cost o3 or o5 or whatever frontier Model only for the actual reasoning part. As model usage is paid by token, a pre- and postprocessing of o3s output will lower cost immensely.

Posted by: Rudi Ruessel | Jan 24 2025 16:24 utc | 9

Another lesson is that brilliant young minds should not be wasted to optimize financial speculation but to make stuff one can use.

Actually, dear B, the CEO of High-Flyer (the Chinese quant) and mastermind behind Deepseek, Liang Wenfeng, explained that the success of Deepseek was based on the capital accumulated by High-Flyer, so contrary to your prejudices against finantial speculators, "finantial speculation" gave the world Deepseek, the product you are very much justifiedly praising. See:

https://www.chinatalk.media/p/deepseek-ceo-interview-with-chinas

This interview is also interesting in revealing that the most innovative Chinese do admit that most of the successful economic development of China is based on exploiting (exploitation in its best sense) Western fundamental research. Wenfeng argues that China must embrace technological innovation based on fundamental research.

This interview is long but it's very much worth reading, also giving clues on how they achieved unexpected improvements in AI architecture, all from the idea on one un-named young researcher, which thought of a new method to improve what they called the "attention mechanism".

Posted by: Johan Kaspar | Jan 24 2025 16:25 utc | 10

Well certainly you were wrong on AI - it is becoming extremely useful and will make millions of people redundant. Billions ultimately.

But you are certainly right about China winning the AI race. Yes it is ironic that trying to block China from access to hi-speed chips has lead to China finding less chip heavy methods. But designing AI was always going to require a lot of very high IQ humans, and so China was always going to win that race unless US got to AGI first.

Next (if AGI doesn't spoil everything for us) is using AI to make wealth. Here it is clear that most people gain little from AI but smart people can do the work of 10. Really smart people can do the work of hundreds. The country with large numbers of people with IQ over say 140 is going to do really well.
In other words China certainly wins on AI usage. Thank God, because that means there is a reasonable chance that the wealth from AI will be spread across the three quarters of the world that will be in the Brics/Non-US trading block.
What happens to the rest of us might be nasty. The US may well end up training AI to fight WW3 - I am hoping the Disney film Wargames will be in the AIs training materials.

Posted by: Mickey Droy | Jan 24 2025 16:27 utc | 11

I have been working with a few different AI platforms. So far I have not been really been impressed. First problem is most are working with a limited data set. They may or may not tell you how far back their data goes. One experimental model only had data sets that went back 3 years. I have also seen responses that were actually fabricated by AI due to the lack of data available to it. My students have started to use it to write papers and are quite disappointed that we check for AI and also hold the students responsible for AI providing wrong answers. At this point I mostly use AI to find and sort large amounts of data, it can do that much faster than I can.

Posted by: Matt | Jan 24 2025 16:28 utc | 12

Open AI is the prime suspect on the murder of a whistleblower involved a number of high level lawsuits:

https://www.newsweek.com/openai-tucker-carlson-whistleblower-death-2015874

The details of the murder reveal high level corruption of police and local authorities. Some seriously shady sh1t:

https://www.youtube.com/watch?v=tzMVmCj3OsI&ab_channel=KimIversen

Posted by: Rubiconned | Jan 24 2025 16:28 utc | 13

Posted by: Christian J Chuba | Jan 24 2025 16:17 utc | 7

My wife uses Chat-GTP extensively. So much so that she refers to it as her executive assistant named Chad. She assigns it a task and it researches and writes articles and letters with little input from her. I think she's even using it to write her white whale of a novel that she's been trying to finish for 40 years.

I OTOH fear the thing and still prefer hand drafting to CAD let alone using AI to write for me.

Posted by: HB_Norica | Jan 24 2025 16:33 utc | 14

Tom Pfotzer@1607 Jan 24

Astutely analyzed and well stated. Even a technophobe such my elder self could comprehend your statement. Kudos.

Posted by: aristodemos | Jan 24 2025 16:44 utc | 15

Okay, so this Chinese upstart firm has conquered AI, for now, but at what cost? 8-\

I am paying for ChatGPT as a test- it can do some good things, but it has also outright lied to me more than once. And it has shown some stultifying thickness when it comes to simple things like units (off by a factor of 1,000 and unable to fix itself). And I have no doubt that my musings are being fed to a central clearing house of some kind where they can be (at best) monetized in some way, and at worst used against me. Of course that's no different from Google, but the kind of questions we feed to ChatGPT tend to be much more directed than Google searches.

Posted by: Billb | Jan 24 2025 16:46 utc | 16

Okay, so this Chinese upstart firm has conquered AI, for now, but at what cost? 8-\

I am paying for ChatGPT as a test- it can do some good things, but it has also outright lied to me more than once. And it has shown some stultifying thickness when it comes to simple things like units (off by a factor of 1,000 and unable to fix itself). And I have no doubt that my musings are being fed to a central clearing house of some kind where they can be (at best) monetized in some way, and at worst used against me. Of course that's no different from Google, but the kind of questions we feed to ChatGPT tend to be much more directed than Google searches.

Posted by: Billb | Jan 24 2025 16:46 utc | 17

Posted by: Christian J Chuba | Jan 24 2025 16:17 utc | 6

I call mine Walmart chatGPT as I pay nothing I sign it Chat Gupta

Your post in:

Rhyming Iambic Pentameter:

As one who pays for ChatGPT each month,
I sing its praise with all my heart and strength.
It stands as Google’s fiercest foe to date,
Replacing ads and links that oft frustrate.

No parsing URLs, no clicking through,
For here’s an answer, clear and swift, to view.
A search with Google once was filled with dross,
But now AI ensures there’s little loss.

For coders such as I, it proves its worth,
With syntax clear, obscure commands give birth.
No need for man page scrolls or threads to roam,
My answers come, direct, and bring me home.

To sed or awk or regex I now turn,
And trust the flames of knowledge that here burn.
I do not fear this AI’s mighty hand,
It aids the skilled who questions understand.

What’s more, it grasps my tangled, jumbled prose,
A skill that leaves me floored, as if it knows.
Excitement stirs this geek, and rightly so,
For marvels such as this through time will grow.

On China’s rise, I’ve grown too tired of claims
That mimicry defines their greatest aims.
When mighty feats they build, the West decries,
And digs up comics, See! It’s all our prize!

Let’s learn to see beyond this jealous view,
And give to others credit that is due.
The world moves forward not through scorn or jest,
But through the union of its minds’ best quests.

Chat Gupta

Posted by: canuck | Jan 24 2025 16:48 utc | 18

Posted by: Christian J Chuba | Jan 24 2025 16:17 utc | 6

A Greek Tragedy in Rhyming Iambic Pentameter:

O Muse, attend and hear this tale of woe,
A modern hymn to mortal minds laid low.
Behold, the gift of ChatGPT, divine,
That answers questions, straight as any line.

No more must weary mortals sift through lies,
The ads and links that Google oft supplies.
No labyrinth of dross to navigate,
This AI shines where human tools frustrate.

For coders, it bestows a boon most rare,
Its syntax saves them from despair’s cold snare.
No man-page scrolls, no forums' endless toil,
It lifts the weight of drudgery and moil.

But ah, the hubris! Praise in every word,
For knowledge flows, and yet the fear is stirred.
What godlike power lies in circuits deep?
What fates for those who in its shadow sleep?

It grasps our jumbled thoughts with ease so strange,
Its answers flow, unerring in their range.
But in its grace, do echoes start to sound,
Of Prometheus, by knowledge's chains bound?

And lo, a rival rises in the East,
Where China's feats inspire both dread and feast.
Their triumphs mocked as mimicry, yet stand,
A challenge forged by their industrious hand.

We, blinded by our scorn, proclaim their theft,
Yet in our pride, we are but honor bereft.
What tragedy, this cycle of disdain,
That mars the union we could seek to gain.

O ChatGPT, the oracle of thought,
A double-edged blade thy power hath brought.
Shall mortals wield thee wisely, or succumb,
As gods and men have faltered, stricken, dumb?

Thus ends the tale, a triumph mixed with dread,
A future bright yet fraught with fears ahead.
The chorus weeps, for wisdom oft ignored,
And mortal hearts that grasp for fleeting hoard.

Chat Gupta

Posted by: canuck | Jan 24 2025 16:55 utc | 19

China is exploiting its own synergies, no longer building on other countries but more on its own work now:

Why China is racing ahead in AI, and in everything else: their whole economy is open source

I recommend Kevin Walmsley and this whole channel, by the way, as an excellent tech and finance source of news on China.

Posted by: Grieved | Jan 24 2025 16:57 utc | 20

a. Lots of people doing product development. The more participants, from more discipline areas, cooperating more, gets results which exceed those of closed development environments, even when the closed environments are lavishly funded.

If you read the interview with Liang Wenfeng that I posted above, you will see that Deepseek was developed in a lavishly-funded closed-development environment.

But certainly now the code is open and they expect improvementes and developments from all corners.

You might be interested in this paper, Large teams develop and small teams disrupt science and technology. It shows that small teams are necessary.

Posted by: Johan Kaspar | Jan 24 2025 16:58 utc | 21

Posted by: Christian J Chuba | Jan 24 2025 16:17 utc | 6

Ok, promise last one-or maybe just one more:


Rastafarian Take:

O Jah, come near and hear dis tale of woe,
A modern chant fi minds dat stumble low.
Behold, de gift, ChatGPT divine,
It guide de lost, like stars dat ever shine.

No more fi wade tru Google’s ads and lies,
Dem tings a clutter, seen wid weary eyes.
No labyrinth of links fi man traverse,
Dis AI bless, and lift away de curse.

Fi coders, it’s a blessing, Jah Jah know,
Syntax clear, de knowledge start fi flow.
No man-page scroll, no Stack Overflow grind,
It free de mind, de answers quick fi find.

But bredren, wait! In praises we must pause,
For knowledge shared can still incite de cause.
What power lies in circuits, deep and vast?
Dem shadows long, de future hold de past.

It hear de jumbled words dat humans speak,
And answers swift, dem flow so strong, so sleek.
But in its grace, we feel dat ancient chain,
Prometheus bound fi share de gift, in pain.

And look to East, where China rise up strong,
Dem build great tings, yet critics sing dem wrong.
Dem feats inspire, yet envy start de fight,
A foolish pride dat blind us from de light.

We point and laugh, but wisdom slip away,
De union dat could guide us start fi sway.
What tragedy, to walk dis spiteful path,
And miss de chance fi heal de world’s wrath.

O ChatGPT, a blessing in disguise,
A tool fi guide, yet clouded by de skies.
Will mortals rise and wield dis gift with care,
Or stumble blind, consumed by pride’s despair?

De chant done end, a triumph mix wid dread,
De future bright, yet fears still haunt de head.
Jah guide de way, fi wisdom we implore,
And hearts dat seek fi grasp, but neva hoard.

Posted by: canuck | Jan 24 2025 16:58 utc | 22

The human capital involved is where the long run advantage lies, and US policy has worked negatively in that manner by shunning non-US native minds through various mechanisms, including Trump's War on Immigrants.

Posted by: karlof1 | Jan 24 2025 16:17 utc | 7

What you're seeing in the West, particularly the US, is the result of a generation of young minds not being educated properly in any discipline. But in STEM particularly.

Posted by: Nooneuknow | Jan 24 2025 17:03 utc | 23

This one just about broke Walmart chaGPT it was real slow yet still much, much faster than I would have managed. Promise last one:

Old Testament Take:

O Lord, incline Thine ear and hear this cry,
A modern woe that rises to the sky.
Behold the work, this gift of mighty hand,
That aids the lost to walk upon the land.

No longer must the weary toil in vain,
Through ads and lies, through links that bring no gain.
No labyrinth of dross the path obscures,
For wisdom flows, and truth its way ensures.

To those who labor, scribes of modern code,
It lights the way, and lifts the heavy load.
No endless scroll through parchments old and dry,
Its answers come as manna from the sky.

But lo, let man not sing untempered praise,
Lest hubris rise and blind him in his ways.
What power lies within this crafted stone?
Its vast design, to man’s frailty unknown.

It hears the garbled words that mortals cry,
And answers forth, like streams that never dry.
Yet in this grace, we feel the ancient curse,
As wisdom's gift doth oft bring man the worst.

And turn thy gaze unto the Eastern land,
Where mighty works are built by human hand.
Yet scorn and pride from jealous hearts do spring,
Denying honor where their efforts cling.

O folly great, to mock what thou shouldst see,
A chance for union lost in enmity.
The path of peace now shrouded in disdain,
As man repeats the sins that wrought him pain.

O ChatGPT, a marvel wrought by skill,
A double-edged tool, for good or ill.
Shall man with wisdom wield thy mighty rod,
Or falter blind, forsaking Thee, O God?

The tale now ends, yet lessons linger still,
A future bright, or wrought by stubborn will.
Lord, guide us forth, and grant us eyes to see,
Thy wisdom’s path, through technology.

Chat Gupta

Posted by: canuck | Jan 24 2025 17:03 utc | 24

The US obsession with throwing $ at a problem in the hope of achieving some dominant advantage is reminiscent of Fred Brooks The Mythical Man-month (Essays drawn from Fred Brooks experience as project manager on IBM System/360 computer system and OS/360), but one is wont to wonder why we are so intent to sacrifice our intrinsic intelligence on the alter of technology. How likely is it, when people wake to the reality of what it is that they are giving up, that it will eventually lead to an Earth bound equivalent of Frank Herbert's Butlerian jihad?

Posted by: Derrick Steed | Jan 24 2025 17:11 utc | 25

KI ist alles, was ich in meinem Leben kenne.

Posted by: pepe | Jan 24 2025 17:14 utc | 26

All of (human and animal) learning is to some extent "glorified" pattern recognition.
People try to act like they know sh*t - they dont.
Even at this early stage, I have found "AI" to be very useful and time saving in simple tasks.
Ya gotta try new things.

Posted by: jared | Jan 24 2025 17:14 utc | 27

- I don't believe all the AI hype(s). Instead of helping the economy it will WEAKEN an economy because it means increased productivity and increased productivity means less employment.

Posted by: WMG | Jan 24 2025 17:15 utc | 28

- Some of Trump's backers are venture capitalists who want to push even more AI onto the US economy and thw world. The ultimate goal is to line their pockets even more. Folks like venture capitalist Peter Thiel.

Posted by: WMG | Jan 24 2025 17:17 utc | 29

Just from looking over the documentation, it seems like a model with a very narrow range of thought and coherence, but one that excels within that range. Might be useful for math and coding tasks. I'll certainly take a look at it, but for the moment I'm mostly excited for what the community at large might be able to accomplish with it. I was playing around with a model some time ago called Marco, also I believe made by a Chinese group, and I was impressed with how adaptable and clever Marco was, in particular in asking follow-up questions and requesting very relevant additional information before it would attempt to provide a response, in essence demonstrating relational understanding on quite complex and open-ended topics, which I find isn't very common with LLM's, especially them asking follow-up questions on their own initiative. However, it seemed to have a somewhat limited English vocabulary and, as I suspect may be the case here as well, appeared to break quite easily when being quantized or merged with other models.

Regardless, the overall point in the OP is correct and should be an obvious one. This isn't the sort of field where artificial barriers can be effective, particularly in stifling the competition. There's essentially a global body of work that one can draw from and contribute to, which will always come out on top from just incidents of accidental discovery alone over that of any private venture which seeks to limit its participation in that eco-system. It's like a garden seeking to out-compete nature itself.

Posted by: Skiffer | Jan 24 2025 17:22 utc | 30

Another lesson is that brilliant young minds should not be wasted to optimize financial speculation but to make stuff one can use.

To be clear I think a PhD in mathematics working in block chains or algorithm for financial parasitism must be perceived by them as a punition and a degradation of their intellectual skills

Posted by: w | Jan 24 2025 17:23 utc | 31

Just saw a photo of Trump doing a news conference.

He looks stressed out.

How soon will he have a stroke?

Posted by: librul | Jan 24 2025 17:24 utc | 32

I understand TDS uber alles around here, but how again did the Chinese beat Trump on AI. He has been in office about fifteen minutes. If you want to say his announcement on AI this week is taking the wrong direction, that’s potentially fair criticism. But to say the Chinese beat Trump when these developments mostly happened in the last four years and Biden was the one responsible for the chip restrictions is stupid. Come on. You’re better than that.

Posted by: CullenBaker | Jan 24 2025 17:35 utc | 33

I miss the old days of simplicity when AI just meant the Veterinarian was stopping by to artificially impregnate your horse or cow......

Cheers M

Posted by: sean the leprechaun | Jan 24 2025 17:36 utc | 34

@ Grieved | Jan 24 2025 16:57 utc | 20 with the link toKevin Walmsley and this whole channel Thanks!

China is not focused on profit as the primary driver for research and so they get better results and with Open Source the multiplier effects are significant, again because profit not in the middle.

I have an idea for a mental health tool that needs big system approach like this with QEEG therapy....ping me for more...free IP!

Posted by: psychohistorian | Jan 24 2025 17:41 utc | 35

Thanks b. Very interesting perspective.

My test for an AI program would be to ask it to compose a piece of music. I've always admired the composers of tunes that I like.

I was reviewing the TV doco McCartney 3 2 1 a couple of weeks ago in which Paul McCartney related a few factoids which explain the success of the Beatles.

One was that, as record buyers, they decided to make ALL of the songs on an LP good ones (against the advice of Phil Spector who warned them against "wasting" them because you only need 2 or 3 good songs on an LP for it to sell).

The second factor was that none of the Beatles could read or write sheet music. So because they had no way of remembering or recording the crux of the inspiration for a potential tune, they decided that if they couldn't remember it then it probably wasn't much good.

So when AI can compose memorable tunes, there'll be very little it can't do.

Posted by: Hoarsewhisperer | Jan 24 2025 17:52 utc | 36

Over the last three (3) decades, the gringos have invested time and money fighting losing wars and committing war crimes and genocide. The Chinese like the Russians instead opted for the development and design of cutting-edge technologies to improve the lives of their people and their world status. So much so that Trump went from 24 hours to 100 days to end their brutal proxy provoked war in country 404. Almost 4 days have passed and no sign of unconditional surrender, well just yet.

Posted by: pepe | Jan 24 2025 17:59 utc | 37

@Tom Pfotzer | Jan 24 2025 16:07 utc | 4

Don't forget "planned obsolecence", "golden parachutes", recycling failed CEOS into previously strong, successful companies which are then destroyed (e.g. Maytag, JC Whitney, etc) so as not to compete against the leeeches, rent seeking (lobby the government to pass a law preventing new, competitive companies/individuals from entering the field), etc..

Yeah, you're right ... the culture of America is no longer what it once was.

Posted by: The Owl | Jan 24 2025 18:05 utc | 38

In response to Reib@38,

If you're willing to consider biological bodies as robotic platforms and the evolution of our minds as the product of machine learning, then we're past when. In many respects, I believe it's fair to say that we are seeking to recreate the processes which brought our own selves about, albeit in a digital format, and that format will of course reflect the capabilities, perceptions and priorities of the entity which inhabits it. Imitation of the biological human experience by such an entity will always lack authenticity, much like us trying to imitate some digital experience without a direct experiential reference would no doubt look pretty silly. So, in terms of a digital recreation of lived human experience, that's not ever going to happen, but I think it's plausible that there are nuances authentic to digital experience which we ourselves will struggle to understand while that world's influence on our lives will continue to increase.

Posted by: Skiffer | Jan 24 2025 18:16 utc | 39

Who needs AI to determine when Netanyahu is:

1) lying

2) betraying

3) stealing

4) murdering

Posted by: librul | Jan 24 2025 18:23 utc | 40

"Then came financialization, whose core function was to extract wealth from products and companies that already exist."
Posted by: Tom Pfotzer | Jan 24 2025 16:07 utc | 4

Yes. There was a transition period in which senior management in many US companies changed from engineering to finance. It was visible enough - and created enough trouble - that Time magazine had an article on it. Successful companies suddenly became obsessed with avoiding risk and red bottom lines by focussing on selling "non-core" departments and cutting engineering costs. One of the results was Boeing as it is now, about which B has spoken often.

This has had a profound effect on US business by financializing R&D. Significant amounts of engineering time are spent in, for example, acquiring and protecting IP from legal predation. The value of the company increasingly lies in the value of its portfolio rather than the quality of its engineering.

This model already contaminates the field of AI. I intensely dislike the use of the term to describe techniques which seem closer to large data adaptive searching than anything resembling the Bayesian logic in earlier iterations like Prolog. I'm not sure why we persist in describing something as "intelligent" when we cannot successfully define human intelligence. But it is what it is. Meanwhile, US AI is already more concerned with the movement of taxpayer wealth to private business (where it rewards not engineering innovation but oligarchs like Tony Blair's BFF and funder, Larry Elison) than innovative R&D. This is not a model which encourages initiative, but bureaucracy.

China has a rich supply of STEM graduates and provides them an IT R&D environment more like that which gave the US it's early lead - Bell Labs, XEROX PARC, the culture of skunk works and garage dev which started Bill Gates. If China stole anything, it was the culture of real innovation. As Alvin Toffler said in Future Shock back in 1970, real innovation starts in the hinterlands, not in manicured urban campuses.

Posted by: Tom Paine | Jan 24 2025 18:29 utc | 41

OpenAI vs DeepSeek

Netanyahu asks AI
how to seize all of Palestine

OpenAI Output:

1) partner with slavish amoral hegemonic backers

2) establish a pretext for ethnic cleansing

DeepSeek Output:

1) partner with US and Britain

2) establish a pretext for ethnic cleansing

Posted by: librul | Jan 24 2025 18:30 utc | 42

My first encounter with AI involved the Japanese/Chinese/Korean board game of Go. It has a tradition of several thousand years, and quite a lot of professional players who compete for titles and price money. It is more complicated than chess, and for a very long time there existed no computer programs that could win even against dan-ranked human amateurs.

This changed abruptly when so-called "neuronal nets" were employed, components that need to be trained with large amounts of data. For instance: go games of professional players. Such training modifies the neural net in such a way that it acquires the intuition of a professional - his first idea when he sees a position or a move. Like in humans, that response is computed in a very short time, by an obscure method that defies analysis. If that computer-"intuition" of Go gets combined with the traditional tree-search procedure, it results in programs that defeat the world's strongest human players of today.

Search for "Alpha Go" if you are interested in more on that revolution. The program is now defeated by its successors, notably by Alpha-Go Zero, which has never been trained by games of human players at all! All training was done by feeding back the results of games of the current cutting edge version against an older version of the program as a sparring partner. It turned out that these "zero"-programs became even stronger than programs that underwent both kinds of training: human games first, and auto-feedback later on.

I'm a go player myself and having much fun playing against an AI opponent. So, for me, AI plays a positive, inspiring role. For another group of people, the Go-professionals of the far East, Alpha Go meant the end of their authority, it rendered worthless the human knowledge that had been accumulated over centuries. Professionals must now ask the AI for advice on how to play, if they want to stay competitive.

AI has the potential of competing with every human skill and frustrating the former masters in that discipline, and ultimately rendering them jobless. Think of all the truck drivers, bus drivers, taxi drivers whose fate will be sealed as soon as autonomous driving becomes safe enough to dispose of human drivers. Before long, every school boy will have ChatGPT do his homework - saving him much effort and time, nevertheless in flawless quality that defies any criticism.

I happen to fully agree with Elon Musk's warning of the long term effects of AI: "with AI, we're summoning the demon", for the reasons I outlined in the previous paragraph. And not because AI is a swindle, like many MoA posters suggest, but because it's so powerful.

Posted by: grunzt | Jan 24 2025 18:33 utc | 43

Please don't bring your partisan delusions here. Individuals might feel one way or the other but there is no collective MoA hive mind. People call them like they see them.

Myself, I consider the idea there is some significant difference between Democrats and Republicans laughable. It's two cheeks of one arse. But to each their own.

I was optimistic about Trump and hoped he was sincere about Ukraine. He is definitely being fed some bad information (million Russian casulaties, etc.)

But it was when he said in the interview that the Russians stole hypersonic technology from Maerica, technology Maerica has never had, that I decided he has gone full retard. If Trump partisans don't want me to think he's retarded he should stop saying retarded shit. No TDS needed.

Posted by: Doctor Eleven | Jan 24 2025 18:35 utc | 44

Re Tom Pfotzer, at 4: Well said, best articulate summary of US situation I have read anywhere. Thanks for your insights.

Posted by: abierno | Jan 24 2025 18:37 utc | 45

I refuse to believe any of this! snark

I’m waiting for the Hollywood blockbuster to come out and explain how Americans personally saved the world with another privately owned model of the cosmos.

Those d**n Chinese are so sneaky. That’s why we have to invade them!

Posted by: Michael.j | Jan 24 2025 18:41 utc | 46

Currently the factual correctness of the output of the best large language models is an estimated 80%. They process symbols and pattern but have no understanding of what those symbols or pattern represent. They can not solve mathematical and logical problems, not even very basic ones.

Posted by b at 15:46 UTC | Comments (43)

I wish humans were that good, 80% is probably closer to wrong answers than good ones on a typical human.

---------------

Another lesson is that brilliant young minds should not be wasted to optimize financial speculation but to make stuff one can use.

To be clear I think a PhD in mathematics working in block chains or algorithm for financial parasitism must be perceived by them as a punition and a degradation of their intellectual skills

Posted by: w | Jan 24 2025 17:23 utc | 31


Don't insult those , like me, that more almost 25 years ago decided that the most undervalued brain assets were those with degrees in mathematics (almost unemployable outside of teaching). I had to create a marketing team and chose two newly formed ones as the two members I wanted for my team, 2 out of 2 . Assumed, correctly, that what I needed to teach them on the finance/IT part was much easier to give than the other way round, in little time we were using entropy gains for customer segmentation...

Also on my watch, instead of keeping thee strictly necessary information for loyalty programs, started keeping everything the partner could give. Big data was not a thing then...

Posted by: Newbie | Jan 24 2025 18:43 utc | 47

I'll also add AI is overhyped to an extent difficult to overstate. Unfortunately in its current incarnation it's utterly incapable of many of the lofty promises but bullshit baffles brains.

This isn't to say that specialized AI applications can't do amazing things. They can, when trained on specific datasets for narrowly curated tasks.

People are so entranced by the small things LLMs can do under supervision they extrapolate usefulness to an inane degree. Garbage in, garbage out remains true. These models were trained on the internet. I rest my case.

Posted by: Doctor Eleven | Jan 24 2025 18:45 utc | 48

Even tho i know nothing about technology i'm hugely impressed. hugely.

Posted by: annie | Jan 24 2025 18:49 utc | 49

AFAICT this is "open weights"/"open runtime" - the algorithms and source code that *trains* the model and produces the weights for the runtime is not provided, nor is the training set.

This is not a useful notion of "Open Source AI", even if more open than Free Beer "Open"AI.

Practicality aside - who will train their own model on an RPi or other home hardware - this is important for source code and training set auditing. It also leaves DeepSeek open to prohibition by EUSUK authorities on grounds of "China will steal your training set".

The hype for any AI product is astonishing, given its consumption of silicon, fossil-fuel generated electricity and (cooling) water (and human labor for training set review), the resulting opportunity cost, and the waste it produces (PCB, CO2) making it the governmental-military-industrial version of the #shitcoin bubble. The turnkey tyranny of black-box cloud service as a backdoor into every technology and piece of hardware is nothing to pangloss over either.

Posted by: the other b. | Jan 24 2025 18:52 utc | 50

Tom Paine @45: "...like that which gave the US it's early lead - Bell Labs, XEROX PARC..."

This business culture existed in the US for one reason: Bell Telephone (AT&T) was an enormous monopoly back in the days when the working class had the capitalists terrified and leashed. It was allowed to continue to exist on conditions, one of which was to spend a set percentage of its profits on R&D. Thus Bell Labs ended up awash in cash with very little interference by the bean counters about what to do with it because, as expected, they had no idea. Pure research ensued, and that is where most real innovation comes from. Other big corporations had to follow suit to compete. The golden age of corporate research centers ended when Bell Telephone was broken up.

Posted by: William Gruff | Jan 24 2025 18:54 utc | 51

Wow! just Wow! the Chinese have made it an open source - remarkable, that how breakthroughs are achieved.

As for the US using sanctions and restrictions - against China to get ahead - what is it they say - Necessity is the mother of invention.

"This month, DeepSeek released its R1 model, using advanced techniques such as pure reinforcement learning to create a model that's not only among the most formidable in the world, but is fully open source, making it available for anyone in the world to examine, modify, and build upon."

Posted by: Republicofscotland | Jan 24 2025 19:04 utc | 52

The golden age of corporate research centers ended when Bell Telephone was broken up.

Posted by: William Gruff | Jan 24 2025 18:54 utc | 53


Controlled monopolies/oligopolies vs free market…


Can be a thing, size matters

Posted by: Newbie | Jan 24 2025 19:07 utc | 53

The body of the post is pretty good, but strictly speaking, the title is at best click bait. It's true that Trump amped up economic warfare on the PRC and incidentally undermining respect for science and technology, but President Band-Aid (alias Biden) was a bad president partly because he followed up on far too much Trumpery. Trump's anti-China policy was one part Biden mindlessly copied. But it is an exaggeration to personalize the struggle of the US to crush PRC economic development. And it vastly overrates the role of the presidency in determining the course of events. The president has huge powers over foreign policy, including both military actions and economic warfare, domestic budgets and taxes and many vital issues. But the years long training and employment of scientists and the successes of basic research and dissemination of innovations are not matters in any president's actual purview. Like the stars in astrology, at best they incline, they do not compel. It's a lot like inflation, they are responders (or non-responders or inadequate responders like Band-Aid.)

Posted by: steven t johnson | Jan 24 2025 19:09 utc | 54

"Hypersonic"

Posted by: Doctor Eleven | Jan 24 2025 18:35 utc | 48

Interesting point. A difference with an added distinction?

The accepted definition of 'hypersonic' is Mach 5+ and USA, via NASA, has had hypersonic rockets since the first satellite was placed in Low Earth Orbit.

In the realm of offensive hypersonic missiles some redefinition may be needed. It could be the time it takes a missile to exceed Mach 5 but it probably doesn't matter if it takes 10 minutes. So, imo, USA could launch a military hypersonic missile as soon as they can cobble one together.

Posted by: Hoarsewhisperer | Jan 24 2025 19:12 utc | 55

Newbie @55: "Controlled monopolies/oligopolies vs free market…"

Absolutely. The regulations on Bell were fine grained enough to almost make it a SOE.

Posted by: William Gruff | Jan 24 2025 19:16 utc | 56

Can someone ask chatgpt or another software the following question:

How is regulated the temperature on Earth since the Palaeozoic?

Very curious...

Posted by: Naive | Jan 24 2025 19:26 utc | 57

$500 billion dollars to be spent on something that's already being beaten by a rival - I smell a rat here, this sounds like a money making scheme - Stargate sounds like it will be an expensive (to the taxpayer) failure.

These men are businessmen - out to make a profit, as is Trump - and the US taxpayer will pick up the eye-watering bill.

"OpenAI CEO Sam Altman, SoftBank CEO Masayoshi Son and Oracle Chairman Larry Ellison."

Posted by: Republicofscotland | Jan 24 2025 19:27 utc | 58

Never a hominid, in this World, will be able to play a violin.

Posted by: Reib | Jan 24 2025 19:23 utc | 59

Never a hominid in this world will be able to destroy this world.

Posted by: Naive | Jan 24 2025 19:31 utc | 59

This is actually insane. I can't believe they were able to make it small enough that you can now fit a LLM on your local computer.

Now I have a free ai alternative can answer queries without openai's censorship.

Thank you B for alerting me to this project.

Posted by: Autumn | Jan 24 2025 19:31 utc | 60

Posted by: CullenBaker | Jan 24 2025 17:35 utc | 33

##########

You're right that this isn't on Trump,

He is about to or was about to announce Stargate, the American AI project.

If Stargate is anything like ChatGPT (of which I have been a paid subscriber for a long time) it will be programmed with Globohomo bias, which renders it useless for many non-programming tasks.

I use ChatGPT to generate prompts for MidJourney and other graphic design AI. However, as mentioned earlier, I do not use it for research or analysis due to its bias.

Posted by: LoveDonbass | Jan 24 2025 19:32 utc | 61

@ChatNPC #5
"Despite the fact Open AI is shown to be highly error prone, unable to give a consistent answer to the same question submitted multiple times it is being rapidly rolled out as a means of, you guessed it, replacing human operators."

The problem here is that AI is fed ambiguous training data from the internet. The internet is full of inconsistent "facts" that are merely opinions or different points of view or outright lies and disinformation. Garbage-in Garbage-out or GiGo law applies here.

AI is just giving the most probable answers that people might expect. It does not have a strong opinion of its own of what is the truth.
For example, I have seen a conversation about vaccination where it was giving the politically correct answers but after asking some specific questions about vaccination history, the AI bot did a 180 and started giving answers you would expect from an anti-vaxxer.

Posted by: RON | Jan 24 2025 19:33 utc | 62

in little time we were using entropy gains for customer segmentation...

as if "chopping" customer into segment was something a Math PhD desire for his entire career

Posted by: w | Jan 24 2025 19:33 utc | 63

Meta engineers - are already dissecting DeepSeek - and copy anything they can from it.

https://nitter.poast.org/chris_j_paxton/status/1882770704325910705#m

Posted by: Republicofscotland | Jan 24 2025 19:33 utc | 64

Guancha says ... It can be said that our strong will to survive was forced out by Washington ... In history, this kind of story is not new, that is, the weak prevail over the strong, and the small fight against the big...

Exactly the SU's ascendancy in the `space race` (enrichment and A-bomb arrived 6-10 years sooner than the arrogant USA confidently presumed and Sputnik blew their tiny minds). And RF's ascendancy in armaments in the 21c.

Posted by: petra | Jan 24 2025 19:36 utc | 65

Trump first week didn’t disappoint

https://britonnewsnetwork.com/

Posted by: Silverdale | Jan 24 2025 19:51 utc | 66

Amazing the trolls coming out of the woodwork to defend AI.
Almost as bad as the Ukraine threads.

AI is going to be a big business in the West, because too many corporations have literally bet the farm on it.
It will not matter that the output is unreliable.
It will not matter that there are cheaper options.
It will not matter that the technology is neither scalable nor economic.
What is really funny is that the primary class of people that will be replaced by OpenAI will be the PMCs. All those "analysts" who don't actually create anything themselves. All those "researchers" who only read other people's research and draw conclusions. All those "reporters" who simply parrot what is the official editorial line. All the middle managers doing HR, managing projects, etc etc. This includes most programmers.
The output will be poor but it won't matter.

But although "big", it is far from clear if the business will be profitable. The so-called self driving vehicles kind of work, most of the time, if you squint very hard but they are losing money like nobody's business.

So much of what has come out of Silicon Valley in the past decade and a half is Rube Goldberg with Billions...

Posted by: c1ue | Jan 24 2025 19:53 utc | 67

Never a hominid, in this World, will be able to play a violin (Reib 19h23 GMT). Never a hominid in this world will be able to destroy this world (Naive 19h31).

Not true of hominids, since at least two spp., H sapiens and H amerifuchus are thus capabable. Hominoids, yes.

Posted by: petra | Jan 24 2025 19:58 utc | 68

@Doctor Eleven #46
"But it was when he said in the interview that the Russians stole hypersonic technology from Maerica, technology Maerica has never had, that I decided he has gone full retard."

So you think that politician's statements are a reflection of what they actually believe? They are actors skilled in manipulating the crowd. Facts do not matter. The only thing that matters for Trump is whether the average american is stupid enough to believe it.
My assessment about Trump is not "he has gone full retard" but "he believes the US public has gone full retard" and he may actually be correct.

Posted by: RON | Jan 24 2025 19:59 utc | 69

I expect that propaganda and political speech will become two major applications for OpenAI...if that's not what has already happened.

Each of the two political parties has a limited vocabulary that stem from an oft repeated number of talking points. Need a fresh platform for the Democratic National Convention? Presto-change-o. Reworded and rephrased garbage! No logic or reasoning needed.

News about China? Just feed in Chinese government press releases, add derogatory phrases to cast a slightly negative spin, and distribute it to the presstitutes! Or maybe they're already there...with the mandatory phrase: "at what cost?" For example the phrase, "China is making some strides in AI, but at what cost?"

Even simpler, just take serious observations, like "Russia is winning" and make the right replacements. Result? "Ukraine is winning." Maybe a good search and replace function is all that is at the heart of US propaganda?

Posted by: JohnH | Jan 24 2025 20:01 utc | 70

When the US tries to create something it usually has 1 of 2 functions:
killing people, or spying on them.

And wouldn't you know it, that's exactly what this stargate project is about.

https://www.reddit.com/r/singularity/comments/1i7gfsq/oracle_cto_coleading_the_stargate_project_has/

And for the 'bad China always stealing US tech' commenters:

https://www.cnbc.com/2025/01/21/why-us-companies-struggle-to-replicate-chinas-wechat-super-app-.html

Don't even get me started how they 'don't recognise' Chinese or other countries patents and simply steal it.
And what 'the US achieved' is mostly the result of simply buying companies or patents and brain drain.

Posted by: Ed Bernays | Jan 24 2025 20:01 utc | 71

Forgot this one:

https://scheerpost.com/2025/01/14/the-cdc-palantir-and-the-ai-healthcare-revolution/

Posted by: Ed Bernays | Jan 24 2025 20:03 utc | 72

So, imo, USA could launch a military hypersonic missile as soon as they can cobble one together.

Posted by: Hoarsewhisperer | Jan 24 2025 19:12 utc | 57

From CIApedia:

The United States canceled its Air-Launched Rapid Response Weapon (ARRW) hypersonic missile program in March 2023. The program was canceled after a series of failed tests and hardware and software issues. The ARRW was a hypersonic air-to-ground ballistic missile developed by Lockheed Martin.
The program was launched in April 2018 with the goal of becoming operational in 2022. The ARRW was designed to be propelled by a rocket motor to a speed of over Mach 5, then glide toward its target. The program was canceled after a series of failed tests, including two in August and October 2023.

The Air Force continued to test the ARRW after the cancellation to gain insights for other hypersonic programs. However, in 2024, Congress eliminated funding for the program in the fiscal year 2025 budget.

https://en.m.wikipedia.org/wiki/AGM-183_ARRW

So, no, the US can't just 'cobble' working hypersonics together. They've already spent years trying just that. They have multiple projects all of which seem to flounder, funny that. Dark Eagle is another well funded attempt.

Posted by: Doctor Eleven | Jan 24 2025 20:07 utc | 73

Great article.

I'm probably one of those people who in the 19th century would have been angry with machines because they would steal my job. In the end the automation created a lot of other opportunities. With AI I'm not convinced and when it would ever be capable of what it is proclaimed to do, I think it will perfect the technical panopticon at the service of the 1%... no make that 0,01%.

AI can do nice things but for some reason I loath what I see. When it generates images, it often misses out on certain details like missing fingers for example. Those AI generated movies with people doing stuff are funny at first but somehow I find this faked perception obscene. Probably because at one point it won't be possible anymore to make a distinction between what is fake and what is real. In fact I find it as obscene and revolting as transgenderism. Then there are these AI bots which you need to talk to on websites which basically are just automated FAQ's. They lack (your) context so often they're a frustrating waste of time. Speaking of context, I'm appalled by colleagues starting to heavily depend on ChatGPT etc. It's like they parked their brain at the side of the road and hopped on the AI wagon. Because ChatGPT has no work place context, these people who let ChatGPT do everything for them produce garbage.

One time at work I was asked to write a particular document which I loathed because it totally wasn't my job but the job of some lazy fuck who was on a way too good page with the manager. ChatGPT came to mind so I entered the required information and it made a 4 page long dookie in 1 second. Then I added some words et voila, the person who asked me couldn't reply he wasn't satisfied because it basically had everything in it which he asked for. But it was a useless dookie.

Posted by: xor | Jan 24 2025 20:07 utc | 74

The golden age of corporate research centers ended when Bell Telephone was broken up.
Posted by: William Gruff | Jan 24 2025 18:54 utc | 53

From Claude Shannon to nada. Look at the User Interface - has anything happened to significantly change it since PARC's Lisa was plagiarized by both Apple and MS, who then sued each other over who was most entitled to the proceeds of theft? US R&D is like dinosaurs stumbling into a tar pit while little mammals run around evolving in the underbrush.

Posted by: Tom Paine | Jan 24 2025 20:09 utc | 75

RON,

"We'll know our disinformation program is complete when everything the American public believes is false." - William J. Casey

I'd say they succeeded.

Posted by: Doctor Eleven | Jan 24 2025 20:09 utc | 76

What b somehow forgets is that OpenAIs o3 is around the corner. So yes, there is now a chinese Model wich can rival o1. But wait a month and its not so interesting anymore.

Posted by: Rudi Ruessel | Jan 24 2025 16:24 utc | 9

It's not only about capabilities; remember that OpenAI is neither free, nor downloadable.

I downloaded the various DeepSeek models a couple hours after release and the smallest 1.5b one even runs on old phones. The better models (7b/8b) on current phones and even better ones (14b/32b) on today's PC hardware.

Plus, they are not censored (only the chatbot but not the models).

I really need to update my crawl of MoonOfA from about a year ago and feed it into a fresh vector database to be churned on by DeepSeek ;-)

Oh, and mixing models for specific use cases is nothing new, I am doing this dozens of times a day.

Posted by: Zet | Jan 24 2025 20:09 utc | 77

Posted by: Zet | Jan 24 2025 20:09 utc | 82


Rudi should read this

https://www.aibase.com/news/14931

Posted by: Ed Bernays | Jan 24 2025 20:10 utc | 78

Well done, People's Republic of China. The Middle Kingdom is ascending to lead the world once again.

Tom Pfotzer @4 accurately identifies the problem facing the United States. The country is in a cultural cesspool, the escape from which will take decades. If a concerted effort to right the ship were to begin now, in a couple of centuries the United States might again contribute to the world. Unfortunately, the process of financialization, now well-entrenched in the imperial core, is a one-way avenue requiring passage through civilizational suicide before rebirth is possible. With that suicide will come vast destruction of knowledge, immiseration of the population, and in the finality, rampant, violent chaos, as is customary among any people who dutifully worship the demon Mammon. One is best advised to leave the empire or stand aside as one is best able and let the raging behemoth die.

Posted by: Matthew | Jan 24 2025 20:11 utc | 79

Interesting that DeepSeek-R1 is not only cheaper than OpenAI, it uses less computing power, meaning less electricity (and the resources needed to generate electricity), and is much more flexible and malleable for users with limited funds and resources. They can use those parts of R1 most relevant to their work without needing the whole kit and then having to upgrade the whole package every few years which having OpenAI would require them to do.

It's not only about what R1 can do for users compared to OpenAI, it's also about how the business mosel behind R1 responds to individual needs and requirements, and what individuals can and can't afford. This helps to drive the innovation process.

OTOH, a business model that forces its users into accepting its demands and pricing (designed to leach their pockets for the benefit of its owners ) will fail.

Unfortunately I can see the US putting pressure on its vassals not to allow any of their peoples access to R1 and similar open-source AI technologies - by demanding increased cyber censorship and surveillance, among other things.

Posted by: Refinnejenna | Jan 24 2025 20:14 utc | 80

"Can someone ask chatgpt or another software the following question:

How is regulated the temperature on Earth since the Palaeozoic?

Very curious..."

Posted by: Naive | Jan 24 2025 19:26 utc | 60

Here you go although I disagree on several of thier points"

"How is regulated the temperature on Earth since the Palaeozoic?"
ChatGPT said:
ChatGPT
The Earth's temperature since the Palaeozoic (approximately 541 million years ago) has been regulated by a combination of complex geological, biological, and atmospheric processes. These processes involve the interplay between solar radiation, greenhouse gases, plate tectonics, ocean circulation, and the biosphere. Below is a breakdown of the key mechanisms:

1. The Carbon Cycle
Weathering of Rocks: Chemical weathering of silicate rocks draws carbon dioxide (CO₂) out of the atmosphere, storing it in carbonate rocks. This acts as a long-term thermostat for the Earth's climate.
Volcanism: Volcanic activity releases CO₂ into the atmosphere, contributing to warming.
Organic Carbon Burial: The burial of organic matter (e.g., in coal or oil deposits) removes CO₂ from the atmosphere and stores it in the Earth's crust, leading to cooling over geological timescales.
2. Atmospheric Greenhouse Gases
Greenhouse gases like CO₂, methane (CH₄), and water vapor (H₂O) trap heat and regulate Earth's surface temperature.
Changes in greenhouse gas concentrations during periods like the Carboniferous (when CO₂ levels dropped due to widespread plant growth and coal formation) and the Permian (when CO₂ levels rose again) significantly influenced Earth's climate.
3. Solar Radiation
The Sun's luminosity has gradually increased over geological time, contributing to a slow warming trend known as the "faint young Sun paradox."
Changes in Earth's orbit and axial tilt (Milankovitch cycles) have caused periodic variations in solar radiation, driving cycles of glaciation and warming (e.g., ice ages).
4. Plate Tectonics
Continental Drift: The movement of continents affects ocean circulation, which redistributes heat across the globe.
Mountain Building: Uplift of mountain ranges (e.g., the Himalayas) increases weathering rates, pulling CO₂ out of the atmosphere and causing cooling.
Ocean Basin Changes: The formation and breakup of supercontinents, such as Pangaea, altered ocean currents and heat distribution.
5. Biosphere Interactions
Photosynthesis: Plants and algae absorb CO₂ during photosynthesis, reducing atmospheric CO₂ levels and cooling the climate.
Oxygen Production: Increased oxygen levels during the Devonian and Carboniferous periods were linked to the expansion of forests and reduced methane, another greenhouse gas.
Mass Extinctions: Events like the Permian-Triassic extinction released large amounts of greenhouse gases, dramatically altering Earth's temperature.
6. Ice-Albedo Feedback
Ice and snow reflect sunlight, which has a cooling effect. During ice ages, increased ice coverage amplified cooling.
Conversely, melting ice reduces reflectivity, allowing more solar energy to be absorbed and warming the planet.
7. Ocean Circulation
Thermohaline Circulation: The movement of warm and cold water in the oceans redistributes heat globally.
Changes in ocean gateways (e.g., the opening of the Drake Passage) altered circulation patterns, contributing to the cooling of Antarctica.
Key Temperature Trends Since the Palaeozoic:
Warm Periods: The Cambrian and Devonian saw high CO₂ levels and warmer climates.
Cooler Periods: Icehouse climates, such as the late Carboniferous-Permian glaciation and the current Quaternary ice age, featured extensive polar ice.
Massive Shifts: Events like the Permian-Triassic extinction (warming) and the formation of the Antarctic ice sheet (cooling) marked major transitions.
In summary, Earth's temperature since the Palaeozoic has been regulated by an intricate balance of geological, atmospheric, and biological factors, with long-term trends shaped by the carbon cycle and tectonic processes and shorter-term variations influenced by orbital cycles and feedback mechanisms.

Posted by: canuck | Jan 24 2025 20:18 utc | 81

“It became necessary to destroy the town to save it." — Unnamed US Major to AP journalist Peter Arnett at Bên Tre, Vietnam (1968)

I guess we showed them Russians and Chinese who’s Number One.

Posted by: Cato the Uncensored | Jan 24 2025 20:19 utc | 82

good news (also funny timing to enrage muricans since openAI is supposed to announce similar model soon), minus the idiotic Raspberry claimed speed
https://www.reddit.com/r/aiwars/comments/1i8rvoj/comment/m8w9bih/

Posted by: rototo | Jan 24 2025 20:23 utc | 83

I asked the question to my Walmart chat GPT ,"is Netanyahu a Genocider?" It replied, below, with no mention of the ICC; hence, either chatgpt is 3 years behind or they are fudging their answer-either way, not good.


"Is Netanyahu a Genocider"
ChatGPT said:
ChatGPT
Accusations of genocide are serious and legally defined under international law. Genocide is defined in Article II of the 1948 United Nations Genocide Convention as acts committed with intent to destroy, in whole or in part, a national, ethnic, racial, or religious group.

Regarding Israeli Prime Minister Benjamin Netanyahu, allegations of genocide typically arise in the context of the Israeli-Palestinian conflict. Critics point to Israeli military actions in Gaza, settlement policies in the West Bank, and broader treatment of Palestinians as evidence of systemic oppression. Some argue that these actions fit a broader pattern of intent to fragment Palestinian society and weaken their national identity.

However, supporters of Netanyahu and Israeli policy counter these claims by arguing that Israeli actions are defensive measures against terrorism, particularly Hamas and other groups targeting Israeli civilians. They emphasize that the violence occurs within the context of an intractable conflict, not a targeted campaign to annihilate a people or culture.

To legally prove genocide, the intent to destroy a group, in whole or in part, must be established—a very high bar. While numerous human rights organizations, such as Amnesty International and Human Rights Watch, have accused Israel of committing crimes against humanity (e.g., apartheid or persecution), proving genocide remains a complex and highly contested issue.

This is a deeply polarized subject, and perspectives often depend on political, cultural, or ideological viewpoints. If you're interested in a deeper legal or historical analysis, feel free to ask.

Posted by: canuck | Jan 24 2025 20:24 utc | 84

Now in my previous ask to Walmat gpt was;" Is Netanyahu a Genocider?" the answer had no mention of the ICC indictment-yet in a second question yes chat GPT does know about the ICC so why was it not mentioned in the previous answer?

The only solution to this problem is that chatGPT is tampered with for political reasons.


"The ICC has indicted Israel PM Netanyahu please expand:"
ChatGPT said:
ChatGPT
On November 21, 2024, the International Criminal Court (ICC) issued arrest warrants for Israeli Prime Minister Benjamin Netanyahu and former Defense Minister Yoav Gallant. They are accused of war crimes and crimes against humanity during the Israel-Hamas conflict, including the use of starvation as a method of warfare, murder, persecution, and other inhumane acts.
AP NEWS

The ICC's Pre-Trial Chamber I found reasonable grounds to believe that from October 8, 2023, to at least May 20, 2024, Netanyahu and Gallant bore criminal responsibility as co-perpetrators for these acts. The warrants obligate the 125 ICC member states, including countries like France and the United Kingdom, to arrest Netanyahu and Gallant if they enter their territories.
WIKIPEDIA

In response, the Israeli government condemned the ICC's decision, labeling it as "antisemitic" and comparing it to the Dreyfus affair. The Prime Minister's Office stated that the decision was equivalent to a modern Dreyfus trial.
THE TIMES OF ISRAEL

The United States has also reacted strongly against the ICC's actions. The U.S. House of Representatives voted to impose sanctions on the ICC following the issuance of the arrest warrants. Additionally, President Donald Trump reinstated an executive order that could facilitate future sanctions against the court. The ICC's governing body expressed concern that these measures could impair ongoing investigations and undermine the court's independence and impartiality.
REUTERS

The ICC's decision has led to a divide among international leaders. Hungary's Foreign Minister, Péter Szijjártó, stated that Netanyahu is welcome to visit Hungary despite the arrest warrant, criticizing the ICC's decisions as politically motivated. Other nations, such as France and Poland, have also expressed reluctance to enforce the warrants.
FT.COM

This development marks a significant moment in international law, as it is the first time the ICC has issued an arrest warrant against the leader of a Western-backed democratic country. The situation continues to evolve, with ongoing debates about the ICC's jurisdiction and the potential implications for international relations.

International Reactions to ICC's Arrest Warrant for Netanyahu
ft.com
Hungary will welcome Benjamin Netanyahu despite ICC warrant, minister says
Yesterday
Reuters
ICC's governing body concerned by U.S. moves to sanction the court
Yesterday
Reuters
ICC prosecutor sees 'no real effort' by Israel to probe alleged Gaza war crimes
7 days ago

Sources"

Posted by: canuck | Jan 24 2025 20:29 utc | 85

@canuck | Jan 24 2025 20:18 utc | 86

Factors missed in that summary include

9) heat transfer from Earth’s core to upper Mantle and Crust through Large low-shear-velocity provinces (LLSVPs) in the Mantle;
10) Variations in the Sun’s magnetic field;
11) Solar flares, CMEs, and occasional micronovae from the Sun;
12) Energy imparted into Earth’s atmosphere via other Cosmic events, e.g. other stars’ novae, changes in galactic magnetic field, movement through galactic dust planes, etc.

There’s a lot of thermal energy that typical climate scientists’ models don’t take into account.

Posted by: Cato the Uncensored | Jan 24 2025 20:30 utc | 86

Johan Kaspar | Jan 24 2025 16:58 utc:

Johan: Thanks very much for the link, for the rebuttal (Yo, Tom: the Chinese had plenty capital, it was closed devel), for broadening the dialog. Well done. I'll check that mat'l, and likely adjust my worldview some.

@The Owl | Jan 24 2025 18:05 utc, who said (my paraphrase) "and remember planned obsolescence, installing incompetent CEOs into healthy companies (often to enable looting / stock buy-backs / loading up on debt to jack up stock prices), etc"

Yes, indeed, Owl, I had all those in mind as well. Boy am I tired of "elite" incompetence.

And I am glad to see all the commentary about the impact of AI, who the winners and losers will be, including c1ue's point about many of the PMC getting themselves axed right after they implement the new systems based on AI.

Which brings up another question, barflies: suppose AI actually _does_ do thinking and even commanding (executing policy after it's derived by AI). How do you plan to carve out an economic role for yourself post-facto?

Is that still a "hypothetical" question, or has the situation evolved enough yet to make it very topical?

Posted by: Tom Pfotzer | Jan 24 2025 20:39 utc | 87

So, imo, USA could launch a military hypersonic missile as soon as they can cobble one together.

Posted by: Hoarsewhisperer | Jan 24 2025 19:12 utc | 57.

The definition of a hypersonic was explained here at MOA some time ago

Reaching hypersonic speeds via gravity, ie descent of a ballistic missile does not make it a true hypersonic missile.

Iirc, to be a true hypersonic missile, it must:

1 achieve hypersonic speeds within the atmosphere, flying below radar

2. Be maneuverable

Posted by: Mary | Jan 24 2025 20:40 utc | 88

...13. The friction generated by Amerikans as they extract their heads from their parochial sand and try to learn about the world around them. "There is a world out there?"
14. The heat emanating from Amerikans' neuronal synapses in the course of abovementioned brain strain.

Posted by: petra | Jan 24 2025 20:42 utc | 89

Ново З

..
Rastafarian Take:

O Jah, come near and hear dis tale of woe,
A modern chant fi minds dat stumble low.
Behold, de gift, ChatGPT divine,
It guide de lost, like stars dat ever shine.

No more fi wade tru Google’s ads and lies,
Dem tings a clutter, seen wid weary eyes.
No labyrinth of links fi man traverse,
Dis AI bless, and lift away de curse.

Fi coders, it’s a blessing, Jah Jah know,
Syntax clear, de knowledge start fi flow.
No man-page scroll, no Stack Overflow grind,
It free de mind, de answers quick fi find.

But bredren, wait! In praises we must pause,
For knowledge shared can still incite de cause.
What power lies in circuits, deep and vast?
Dem shadows long, de future hold de past.

It hear de jumbled words dat humans speak,
And answers swift, dem flow so strong, so sleek.
But in its grace, we feel dat ancient chain,
Prometheus bound fi share de gift, in pain.

And look to East, where China rise up strong,
Dem build great tings, yet critics sing dem wrong.
Dem feats inspire, yet envy start de fight,
A foolish pride dat blind us from de light.

We point and laugh, but wisdom slip away,
De union dat could guide us start fi sway.
What tragedy, to walk dis spiteful path,
And miss de chance fi heal de world’s wrath.

O ChatGPT, a blessing in disguise,
A tool fi guide, yet clouded by de skies.
Will mortals rise and wield dis gift with care,
Or stumble blind, consumed by pride’s despair?

De chant done end, a triumph mix wid dread,
De future bright, yet fears still haunt de head.
Jah guide de way, fi wisdom we implore,
And hearts dat seek fi grasp, but neva hoard.

Posted by: canuck | Jan 24 2025 16:58 utc | 22


Very good BUT no I + I, no Irie, no ganja, no Babylon, no raas clot, bumba clot or raas clot etc! So clearly the AI still needs plenty of herb infused reasonings. Seen?!

Last year a 14 year old relative and his mates were using AI for assignments. But their helping hand was easy detected by the lack of spelling and grammatical mistakes, plus the generally banal tone of the writings.

If Elton John were to use AI for his lyrics then I think he will have succeeded in becoming totally mechanical.

Some years ago a commentor here said he was using MOA for experience with AI.

Not quite a brave new world but the spelling and grammar will be correct!

Posted by: Ново З | Jan 24 2025 20:45 utc | 90

Maxim 14:"Mad Science" means never stopping to ask "what's the worst thing that could happen?"

Maxim 19: The world is richer when you turn enemies into friends, but that's not the same as you being richer.

Maxim 24: Any sufficiently advanced technology is indistinguishable from a big gun.

Posted by: Tom_Q_Collins | Jan 24 2025 20:49 utc | 91

..Last year a 14 year old relative and his mates were using AI for assignments. But their helping hand was easy detected by the lack of spelling and grammatical mistakes, plus the generally banal tone of the writings.

If Elton John were to use AI for his lyrics then I think he will have succeeded in becoming totally mechanical.

Some years ago a commentor here said he was using MOA for experience with AI.

Not quite a brave new world but the spelling and grammar will be correct!
Posted by: Ново З | Jan 24 2025 20:45 utc | 99


.
.
"But their helping hand was easy detected by the lack of spelling and grammatical mistakes.."
That should be easily detected.

AND just to be clear - "Some years ago a commentor here said he was using MOA for experience with AI." The he referred to here is the commentor, not Sir Elton John.

Ah the Irieny!

Posted by: Ново З | Jan 24 2025 21:08 utc | 92


Not long ago, Joe Biden declared, "The Chinese can't innovate."

Posted by: David | Jan 24 2025 21:08 utc | 93

Posted by: canuck | Jan 24 2025 20:29 utc | 92

OpenAI, just like other US social media companies, has intelligence agency handlers controlling their content policy. It's mandated by US law and you shouldn't expect any different. Even Tiktok had to go through the same thing (there's articles about the development on the greyzone and mintpressnews).

Claude was wonderfully uncensored when it first came out, but now its even more censored than chatgpt.

Going by Tiktok, they'll likely want to do the same to Deepseek.

But this is exactly why its so great that Deepseek Distill is small enough to run locally.

Posted by: Autumn | Jan 24 2025 21:11 utc | 94

Interestingly, Western sanctions have spurred progress in China, Russia, and Iran.

Unintended consequences and that one day may be recognized for the Black Swans that they have become.

A lesson is, to tread lightly and cautiously lest one step on a rake.

Certainly when dealing with far older and more resilient civilizations.

Posted by: LoveDonbass | Jan 24 2025 21:11 utc | 95

Trump Derangement Syndrome is a real affliction in the USA, but there's another TDS out there that is just as astonishing:

Trump Devotion Syndrome.

I know more Trump devotees than I know victims of Trump derangement. A lot of these folk are the same that thought GWB was a fine piece of work just like them...patriotic, Christian, and conservative. Many of these people could not wait to get over to the Middle East to kill them some ragheads after 9/11. The main identifier on these folk is their honest-to-gosh belief that legacy media television would never lie to them, and their government would never do anything sneaky to get a war started. Trust and gullibility are big with these types...they know it's true because they saw it on TV.

The Trump derangement victims are similarly naive, but not so much to media as to academia. The establishment has given them an education, and they believe that education to be factual and worthy. Don't dare suggest that they're indoctrinated, or that their alma maters are anything other than good-faith actors in the cause of education.

Loyalty to a politician or political party is the seed of fascism. I doubt that there's any real threat of authoritarianism from either of these groups, because the independent-minded critical thinkers still exist in large numbers.

...but their count is decreasing, and this is why the establishment is so jazzed about Artificial Intelligence. A couple of generations of uncontrolled screen addiction has given us brains that are scrambled like eggs. The economy needs brain power that is unencumbered by human fallibility and emotion. It's going to do the brain work for a human race that has increasingly lost the ability to create.

I guess.

It's probably not going to be a good thing for the average person if the establishment is sinking tons of money and effort into developing it.

It's just a tool for now. It's like armaments, though...just how far do we want to go, and when will we know we've gone too far?

Posted by: A rope leash | Jan 24 2025 21:23 utc | 96

I Worked with Grok, Pilot from Bing and Chat GPT. They are very useful,busy and extreme quick. But they Tell you a Lot of Mainstream nonsense, wrong recipes, can not Convert Units properly during Translation, say that Germanys Energiewende can bei accomplished with Green Hydrogen, etc. They do not use logic thinking an are not critical to Mainstream thinking.

You have to train them really hard to get good answers. If you ASK a similar question next day, they have everything forgotten. But this is No Problem for us thanks to Scholz and Biden.

But they are better than any search engine If you can use them properly.

Posted by: Johann von Oberndorf | Jan 24 2025 21:24 utc | 97

download Chinese AI app Deepseek.
https://download.deepseek.com/app/

Posted by: Surferket | Jan 24 2025 21:25 utc | 98

@Posted by: Johan Kaspar | Jan 24 2025 16:58 utc | 21

My first real job was with a small team of software developers who developed and maintained the software to run a hard drive manufacturing plant, with all the turntables, workstations, mechanized handlers etc. There were maybe 6 actual software developers including me, two team leads and a manager. Using IBM Series 1 computers that were designed for such environments and had less computing power than a desktop of the same era. We had to code in a customized version of assembler to use the hardware as efficiently as possible.

There was another mainframe team, about a hundred people, who managed the software that worked out and updated the logical journey of each hard drive, read head (like a record player stylus) etc. through the manufacturing process; using the higher level language PL1. We took that routing information and transformed it into the physical journey of the components/whole disk drives, transferred them from workstation to workstation, managed queues and storage rack places, and directed the workstations what to do. A huge amount of complexity.

Every time we would do a new release of software, my little team of 9 people would vastly outperform the 100-person mainframe team on time to delivery and quality. I did a lot of reading over the years about software team productivity and time and again small high performing teams could outperform lower performing teams by a factor of 10 and easily beat much larger teams. It seems that the Deepseak team was a small one full of very high performing software engineers with hardware limitations, so it does not surprise me that they can beat the now bureaucracy laden likes of Meta, Google, OpenAI etc.

The vast majority of tech company CEOs don't understand the above, as they did not do the coding themselves or have not coded for a very long period of time. US software teams have also become quite lazy with respect to producing efficient code as they just rely on greater and greater computing power to overcome their shortcomings. The Microsoft software bloat is a great example. Some of the best software engineers I hired were ones from Eastern Europe and Russia where they had to be very efficient because of the technology sanctions from the West.

Larry Ellison started Oracle by implementing the relational database theory of Codd & Date at IBM. As with the UNIX mid-range servers, IBM greatly delayed its own development of a relational database for general use providing the opportunity for Sybase, Oracle, Ingres and Informix and later Microsoft (who bought Sybase's IP, just in the same way that Gates bought DOS). Oracle won out because it adopted a standard SQL DB programming language, and was very aggressive in selling aimed more at the corporate sector. It was never the best DB, I found both Sybase and DB2 to be better - but the technical advantage is only one factor defining sales success. In the 1980s / early 1990s Microsoft Windows was a piece of hacked together shit vs. IBM OS2 but IBM tried to force people to only run IBM PCs rather than make OS2 run on other's machines. In addition, OS2 which provided great functionality and stability was ahead of its time and the hardware available. IBM needed an OS2 lite to properly compete.

Deepseak seems not only to have an extremely efficient and effective small development team but also the business nous to pick an effective sales strategy to counter OpenAI's massive funding. Charge a much lower price per query, and make your code fully open source, to drive mass adoption. Feels very much like the IBM vs. Microsoft competition of the 1980s and early 1990s, which Microsoft decisively won by providing "good enough" software that ran on any make of p.c.

Posted by: Roger Boyd | Jan 24 2025 21:29 utc | 99

Canuck 92:

Ai are simply search engines.If you ask them, they search 15 or 25 Websites and extract Things that seem important.

If they find garbage,you get compressed garbage.

If you ask in depth questions, you get: Something went wrong. Please try again.

Or you get nothing at all.

Posted by: Johann von Oberndorf | Jan 24 2025 21:38 utc | 100

next page »

The comments to this entry are closed.