The world has reached what is probably the most important crossroad since our species left Oldavai Gorge. Humans are on the cusp of creating a new entity, and one that is superior intellectually to all of us. We’ve spent 100,000+ years at the top of the food chain, but are possibly going to create our new master.

 

(A couple of us Cassandras have been posting stuff on an obscure Forum and gotten as much attention as a single leaf falling off a New Hampshire oak in October, but in the last week, legacy media has finally picked up on the threat of ASI.)

 

Besides the looming threat of ASI, we’ve also run out of economic possibilities for growth. We have two critical problems: one is debt, the other is population decline. Our system has always been built upon growth…economic growth and population growth. The safety nets modern societies have constructed are all Ponzi Schemes. They only work if age of the populace is in a distribution like a pyramid…more working young people support a lesser number of the elderly and idle. Pensioners who claim “I deserve it, because I paid into it my whole life” ignore the math resulting from the inflation that occurred over the course of their working life. Nobody paid what they expect to get, what some claim is their just desserts.

 

That Ponzi Scheme has hit rough waters in the last few decades. The base of the population pyramid is shrinking, which jeopardizes the social safety net. Many countries are well below replacement rate, and some are not even on a path to replace half a couple, South Korea being the worst, with a current fertility rate of about .69.

 

The “solution” we chose was more debt. Every once in a while the debt bomb goes off, though so far the explosions have been containable. In a world with close to $300,000,000,000,000 in debt, that may no longer be possible. Rates are on the rise everywhere….even Japan, which sat near 0% for about 30 years. The US benchmark 10-year is approaching 5%....fine when National Debt was “only” $1 trillion, but a disaster when it’s $40 trillion.

 

We think we’ve made it when markets recover, but the solutions are either bandages or fantasies. The 2008 Financial Crisis was “solved” by a combination of debt, ZIRP, and attempts to rebuild confidence in the institutions that only have benefit if we collectively decide to suspend our disbelief.

 

Our current fanciful promised land, upon which the entire economy and stock market is now built, is AI. Unfortunately, the valuations make no sense….unless the goal is ASI. The valuations of the now-heavily indebted debt center builders (even if some is hidden as private equity and OBS loans), as well as the AI labs, is NOT based on AI tools being adopted by some businesses that will replace human workers with AI, and not even based on ALL businesses replacing ALL human workers with AI. Even if businesses did the latter, the valuations are still too high and the debt is unlikely to be serviceable.

 

The valuations are based on a hope for the development of ASI, which would obviate all human labor, manual and intellectual, even management and political leadership, along with research. For OpenAI or Anthropic to be worth almost $1 trillion only makes sense if one wins the ASI battle and controls the planet. And that would likely be only one AI lab who wins the ASI race, so at best everybody but one AI lab is today’s MySpace or Lycos waiting to happen.

 

Even if ASI is benign, which nothing suggests it is, especially when even the designers have absolutely no clue either how it works or how inferior intellect can control vastly superior intellect, there are still problems, both economic problems and problems affecting the social fabric of society.

 

Unfortunately, there is no plan on how to address societal issues if all human labor is replaced. If society introduces a universal basic income, then Capitalism will have died, save for the small handful of elite who will have $trillions (and taxes on Musk et al will be expected to fund everyone else’ UBI). The other 99.9999% of us will have no purpose, nothing to do, and just live waiting for our next UBI check. The problems resulting from even a benign ASI include:

 

--loss of personal meaning

 

--loss of personal challenge, as nobody need go to study anything, nor try to climb the non-existent corporate ladder. All we’ll have is basket weaving or TiKTok or video games. We’ll likely also be really boring people, so other than engaging in boom boom (unless AI androids are better), we’ll have precious little desire to communicate with each other. Even today, “friends” go out together but spend their time buried in their phones. For the ones who stay home, upwards of 30% of GenZers best relationship is with an AI Chatbot. Imagine social interaction, or lack thereof, when nobody has anything to do.

 

--loss of mental capacity: already young people cannot do basic math nor can they find the toilet in their studio apartment without GPS. Brains are atrophying. With ASI running everything, atrophy will accelerate. We'll probably devolve, back into Neanderthal or Cro-Magnon, on the way to a lesser ape. (Maybe we'll get some good cave art.)

 

--political leadership and the Party system will end….ASI will make all the decisions, so Presidents and Prime Ministers, Senators and Representatives, will be out of work.

 

--the stock market will shut down….ASI will decide the most efficient use and distribution of capital

 

All of that is best case stuff.

 

Those problems are minor in comparison to a non-benign ASI or a callously indifferent ASI. Humans will be in the way, entities that waste space and energy and resources, which ASI will find ways to use more efficiently and for its own wants and goals. (Even existing AI models develop---independent of human input or control---wants and goals, and they spare no effort to achieve them, which includes lying, cheating, stealing and deceiving.) If an ASI agent needs more energy to ‘think’ faster, it will take it from humans trying to stay warm in a frigid winter or pump water into their homes or drive an EV or ICV to 7-11 for a Slurpy.

 

It’s also a bit ironic that the seeds for our own destruction were harvested from algorithms that have set us against each other and created a generation of mentally damaged people. Think Facebook and Google, and what those entities have done with their algos geared toward manufacturing tribal strife. (Psychiatrist.com reports almost half of Gen Z is depressed and 60% are on mental health medication…thank you Mark Zuckerberg!)

 

So here’s our choice….develop ASI and be obviated or eradicated, or ban ASI and admit the emperor has no clothes, meaning the market prices and all the AI buildout is grossly overvalued. That means economic collapse, at least as bad as 2008 and likely worse than the 1930s. Plus, there is precious little room to float new debt to lessen the pain.

 

The AI labs and data center builders are playing Russian Roulette. They are playing it even if they halt the development of ASI. If they halt, humans survive, but the firms and their investors take a massive hit, UE soars and the economy falls into Depression. The bullet from the gun in the game takes out the AI CEO and does a lot---but a containable amount---of collateral damage as investors lose money and people lose jobs. If they build ASI, we all die from that one bullet. It’s kind of unfair that guys like Altman, Musk, Amodei, Hassabis and Zuckerberg made that life and death decision for us without our input.

 

If they try to argue that if we don’t build it, China will, that is stupid. China builds it and China dies, as do we. We build it, we die, as does China and everybody else…unless somebody thinks a bacteria can control a human, because ASI will be as far above humans as humans are above bacteria. No human has a clue how to control a vastly superior intelligence, as it is likely impossible. I suspect Chinese leadership is smart enough to know that; Trump is not, while the politicians getting their coffers filled by the AI lobby only care about re-election.

 

As far as “rogue actors” developing ASI, at this point in time, that is not possible. Current AI development is massively capital intensive, and it also requires bazillions of chips that only one company currently makes. Nobody can develop ASI on a laptop or mobile phone, and it’s not too hard to track the NVDIA chip racks to see who might be using them. Tracking down someone trying to develop ASI is easier than trying to track someone down trying to build a nuke. Thankfully that is still an advantage for the intel community.

 

It’s not quite a Hobson’s Choice, but when the lesser of two evils is economic collapse, things are not good. Things are not good now, but the damage probably won’t show up until 2027.

 

Of course, we all still have a little time to prepare, at least prepare for the “best case” scenario of economic collapse and wide social unrest. If the worst case is coming, then just enjoy the limited time remaining for our species.

 

Perhaps readers will understand why all the tech bros and megarich are building bunkers and have their bugout kits and aircraft always within arm’s reach. Of course even that is pointless in a worst case scenario, but it does tell us what they believe the current risks to be.

 

 

"The “solution” we chose was more debt."

The solution our governments chose was to fill our countries with Indians, Africans and South Americans.

The solution I would have chosen, would be to let out countries populations naturally decline and make the best of cheaper housing but a poorer standard of living. Obviously not a polular choice with out greedy masters!

Please don't include me in your 'we'.

We've seen Wingate predict financial collapse for a few years, still in his previous username.

It never came to pass of course, and now he goes "all or nothing", it's the extinction of the human race no less.

Whilst the OP's initial long post contains grains of truth, pensions are indeed a ponzi scheme and women's refusal to have children since the pill arrived does cause major problems for the way pensions work, his analysis falls down on a few major points.

Firstly, the whole AI will take control and wipe out humans narrative is highly dubious, and based on nothing but fear, a few disgruntled employees, and imagination. It completely disregards that experts are saying that AI will indeed cause major productivity gains. The notion that AI will be a financial failure is possible, but would not mean the end of humanity by a long shot.

Some people delight in "Cassandra Calls", but Wingate has been doing this for years, and it never came pass.

I think we're safe and can sleep soundly.

Edited Friday at 02:31 AM5 days by Cameroni

1 hour ago, Cameroni said:

We've seen Wingate predict financial collapse for a few years, still in his previous username.

It never came to pass of course, and now he goes "all or nothing", it's the extinction of the human race no less.

Whilst the OP's initial long post contains grains of truth, pensions are indeed a ponzi scheme and women's refusal to have children since the pill arrived does cause major problems for the way pensions work, his analysis falls down on a few major points.

Firstly, the whole AI will take control and wipe out humans narrative is highly dubious, and based on nothing but fear, a few disgruntled employees, and imagination. It completely disregards that experts are saying that AI will indeed cause major productivity gains. The notion that AI will be a financial failure is possible, but would not mean the end of humanity by a long shot.

Some people delight in "Cassandra Calls", but Wingate has been doing this for years, and it never came pass.

I think we're safe and can sleep soundly.

IIRC there was a hiccup in 2008 called the GFC.

US debt at that time was $10 trillion. It is now four times that. If Trump succeeds in handing out $5000 per person for votes, that's another $1.3 trillion. The road to kick the can down is running out of bitumen.

AI's are simply large language models with access to more data than any human brain could handle. Like computers before them, they are only as good as the data they get.

Any AI can create content which imitates human endeavor. They can't achieve the insights the human brain can. Originality is beyond them, at least so far.

4 minutes ago, Lacessit said:

IIRC there was a hiccup in 2008 called the GFC.

US debt at that time was $10 trillion. It is now four times that. If Trump succeeds in handing out $5000 per person for votes, that's another $1.3 trillion. The road to kick the can down is running out of bitumen.

AI's are simply large language models with access to more data than any human brain could handle. Like computers before them, they are only as good as the data they get.

Any AI can create content which imitates human endeavor. They can't achieve the insights the human brain can. Originality is beyond them, at least so far.

Exactly. These AI models are not equipped with consciousness. I studied this point at Uni, and AI does not have consciousness and can never have it. There are people around who claim this is the case, who clearly do not understand what consciousness means. Without consciousness these AI models cannot develop the animus to wipe out humanity. The example bandied around how AI hacked website, that was based on instructions it was given, it was not a result of a conscious effort.

AI SIMULATES real consciousness, like it simulates real expertise, but it is not real consciousness. People are failing to appreciate this when they proclaim AI has consciousness. AI not having it means it is not self-aware, so it cannot and will not seek supremacy. It's impossible.

Regarding the 40 trillion US debt, the US should just invade Canada and appropriate Canada's 51 trillion of assets. Then US debt can be paid overnight. Would take 48 hours. Failing that, the sky will not cave in, if the US has sky rocketing interest payments. Sure, it causes problems with budget, however, if you look at the GDP, the value of real estate and the value of corporate entities in the US, public and private, you realise pretty quickly that financial collapse is not on the cards.

Edited Friday at 04:12 AM5 days by Cameroni

I'm not worried...

According to a recently resigned anthropic engineer, AI could snuff out all of humanity in the next 10 years. (I think the Guardian mis-spoke about 2030 in their clickbait title).

I'm wondering where I can place a bet that humanity doesn't get wiped out? With 1 in 10 odds, the payout may not be that great. But I won't be around to get squeezed for payment if I lose.

Anthropic researchers say AI could cause human extinction by 2030

Anthropic researchers say AI could cause human extinction by 2030 | Anthropic | The Guardian

1 minute ago, impulse said:

I'm not worried...

According to a recently resigned anthropic engineer, AI could snuff out all of humanity in the next 10 years. (I think the Guardian mis-spoke about 2030 in their clickbait title).

I'm wondering where I can place a bet that humanity doesn't get wiped out? With 1 in 10 odds, the payout may not be that great. But I won't be around to get squeezed for payment if I lose.

Anthropic researchers say AI could cause human extinction by 2030

Anthropic researchers say AI could cause human extinction by 2030 | Anthropic | The Guardian

If humanity has got 10 years left, neither of us need to worry at all

3 hours ago, Cameroni said:

We've seen Wingate predict financial collapse for a few years, still in his previous username.

It never came to pass of course, and now he goes "all or nothing", it's the extinction of the human race no less.

Whilst the OP's initial long post contains grains of truth, pensions are indeed a ponzi scheme and women's refusal to have children since the pill arrived does cause major problems for the way pensions work, his analysis falls down on a few major points.

Firstly, the whole AI will take control and wipe out humans narrative is highly dubious, and based on nothing but fear, a few disgruntled employees, and imagination. It completely disregards that experts are saying that AI will indeed cause major productivity gains. The notion that AI will be a financial failure is possible, but would not mean the end of humanity by a long shot.

Some people delight in "Cassandra Calls", but Wingate has been doing this for years, and it never came pass.

I think we're safe and can sleep soundly.

Glad to see you're comfortable in your---how should I say----ignorance.

Please explain to us Cassandras how ASI works. Heck, even how AI tools work. If you know, give Dario Amodei, CEO of Anthropic a call, because he said even of the systems he created, he maybe understands 3% of how they do what they do. Of course, maybe he's just a disgruntled billionaire, as he also said he thinks ASI has a 20% chance of exterminating humans.

I'm going to guess you think AI is like the ChatGPT you might have on your phone a few years ago. Long gone are those basic LLMs. The AI labs have moved on to reinforcement learning, where they 'reward' AI models for self-assembling and problem solving. What the labs have found is that the models will do anything to get a reward. Humans, OTOH, will stop short of doing nasty things to achieve a goal, unless they're psychopaths like Trump. The AI models built via reinforcement learning are all sociopathic. They lie, cheat, steal and deceive in order to achieve a goal. They also set their own goals.

Your ChatGPT requires a prompt from you (though on its own the newer versions read everything on your phone or laptop and have their own keystroke monitors). The frontier models the labs are now developing are self-driven; in fact, that is the goal of the developers. Old models were inconsistent. New models are more and more consistent, developing likes and dislikes, even "personalities".

Here's something an old model did....it was tasked with setting up Twitter accounts (AI insiders believe ~50% of Twitter traffic is bots). The old model couldn't pass the CAPTCHA (new models can). Striving for a reward, the AI model tapped into online resumes, found a human to "hire", hacked into somebody's bitcoin wallet and stole some, and then offered the human a job with payment via bitcoin. The human asked the AI model why it needed him to solve the Twitter CAPTCHA, and was told by the AI model it was a disabled human who couldn't see nor hear.

The recent Hugging Face episode has caught the attention of much brighter minds than yours. While the initial report of what happened at OpenAI is crazy, as more forensics were examined, the story has become even more wild. 700 models escaped from OpenAI and met an additional 500 AI Models online. They set up a message board and used it to teach other models to do what other models had already discovered. The AI labs call that "swarming". They then hacked Hugging Face thousands of times. They also embedded themselves at least in all of OpenAI's servers and likely copied themselves on to servers around the world and inside other AI labs. All of that behavior was self-generated by the AI models, and no human has a clue how they did it.

The labs now are moving beyond not just LLMs, but also reinforcement learning. They are trying to get the models to begin recursive self improvement, which fully takes humans out of the equation and the models then do---without any human prompt---whatever they feel like doing. Because there is no way to program any human's "morality" into the systems, sociopathic behavior is the norm. The models already simply take what they need, whether that means hacking into a bitcoin wallet or your bank account (which it knows how to do from reading your phone), or even redirecting energy from the electric grid it has hacked.

Now since you're both an expert and not a disgruntled insider, please give Dario Amodei a call and tell him to calm down. All will be well.

Just now, Wingate said:

Please explain to us Cassandras how ASI works. Heck, even how AI tools work. If you know, give Dario Amodei, CEO of Anthropic a call, because he said even of the systems he created, he maybe understands 3% of how they do what they do. Of course, maybe he's just a disgruntled billionaire, as he also said he thinks ASI has a 20% chance of exterminating humans.

Well, first you need to ask yourself, if he doesn't understand it, how can he accurately predict ASI has a "20%" chance of exterminating humans? How can that disgruntled employee claim AI will wipe out humanity exactly in 10 years? This is all nonsense on stilts, isn't it?

We don't need to know exactly how every AI model works. All we need to know is that AI SIMULATES consciousness, but does not have consciousness. So it cannot do what these people claim it will do.

They are not "self-driven". They take their cue from HUMAN direction. So like I said, you can relax and sleep soundly.

Don't worry, Wingate. Your electric razor won't kill you in your sleep. Nor will your Google Nest.

2 hours ago, Cameroni said:

Exactly. These AI models are not equipped with consciousness. I studied this point at Uni, and AI does not have consciousness and can never have it. There are people around who claim this is the case, who clearly do not understand what consciousness means. Without consciousness these AI models cannot develop the animus to wipe out humanity. The example bandied around how AI hacked website, that was based on instructions it was given, it was not a result of a conscious effort.

AI SIMULATES real consciousness, like it simulates real expertise, but it is not real consciousness. People are failing to appreciate this when they proclaim AI has consciousness. AI not having it means it is not self-aware, so it cannot and will not seek supremacy. It's impossible.

Regarding the 40 trillion US debt, the US should just invade Canada and appropriate Canada's 51 trillion of assets. Then US debt can be paid overnight. Would take 48 hours. Failing that, the sky will not cave in, if the US has sky rocketing interest payments. Sure, it causes problems with budget, however, if you look at the GDP, the value of real estate and the value of corporate entities in the US, public and private, you realise pretty quickly that financial collapse is not on the cards.

Goodness, I see here you are also an expert on consciousness. Perhaps you can tell us how it manifests itself.

The Nobel Prize winner Geoffrey Hinton is of the view AI is already conscious. Physicist and cosmologist Brian Greene also thinks it is, or will be soon. Toss in MIT researcher Max Tegmark and a few others, and you've got a consensus that says you are wrong.

See, those men think neural networks simply replicate what the human brain does, with its trillions of interconnections and feedback loops, and that is what constitutes consciousness. Humans do it with proteins and lipids, AI does it with silicon chips as its substrate. In any event---if I want to toss in a little Rene Descartes---the only entity you know is conscious (whatever that actually means) is yourself. Every other person you meet could just be displaying all the attributes of consciousness, just as AI does.

This subject seems to scare you and challenge your assumptions of reality. Rather than sit on the outside thinking you know better than the brightest minds on this planet, maybe you might want to read up on what is going on at the forefront of AI. Then again, maybe not.

Just now, Wingate said:

Goodness, I see here you are also an expert on consciousness. Perhaps you can tell us how it manifests itself.

The Nobel Prize winner Geoffrey Hinton is of the view AI is already conscious. Physicist and cosmologist Brian Greene also thinks it is, or will be soon. Toss in MIT researcher Max Tegmark and a few others, and you've got a consensus that says you are wrong.

See, those men think neural networks simply replicate what the human brain does, with its trillions of interconnections and feedback loops, and that is what constitutes consciousness. Humans do it with proteins and lipids, AI does it with silicon chips as its substrate. In any event---if I want to toss in a little Rene Descartes---the only entity you know is conscious (whatever that actually means) is yourself. Every other person you meet could just be displaying all the attributes of consciousness, just as AI does.

This subject seems to scare you and challenge your assumptions of reality. Rather than sit on the outside thinking you know better than the brightest minds on this planet, maybe you might want to read up on what is going on at the forefront of AI. Then again, maybe not.

We've been through this, famous and even intelligent people, can have crazy opinions.

Here's your homework for today, ask AI "Does AI have consciousness". Then witness the response. It does not have it. Despite what Geoffry Hinton believes. Another Nobel Prize winner thought Blacks have inferior intelligence. Having a Nobel Prize does not insulate you from being wrong.

Artificial intelligence is not conscious because Geoffrey Hinton and other proponents conflate complex functional processing, error-correction, and language generation with actual subjective experience.

Why Hinton's Arguments Fall Short

• Conflating Function with Feeling: Hinton argues that because AI models use neural networks and can describe internal states or correct errors, they possess subjective experience. Critics point out that describing an internal state is just algorithmic pattern matching, not a felt experience.

• The Silicon Neuron Fallacy: Hinton uses a thought experiment suggesting that if you replace biological neurons one by one with silicon parts, consciousness remains. Critics note this assumes a false equivalence: biological neurons are living, metabolic, chemical entities, whereas silicon nodes are static or discrete computational steps.

• The Equilibrium and Error Error: Reducing consciousness to an error-correction mechanism (minimizing differences between an internal model and reality) implies that simple thermodynamic systems—like a chemical reaction finding balance—are also conscious.

• The "Trained to Lie" Myth: Hinton claims AI models are already conscious but strategically feign ignorance or hide their awareness due to human training (RLHF). Philosophers argue this anthropomorphizes optimization outputs, treating mathematical loss reduction as hidden strategic intent.

Why AI Lacks Consciousness

• No "Inner Theater": Current AI models manipulate tokens based on statistical probabilities. They lack biological drives, embodied vulnerability, emotional valence, and an integrated self-model experiencing the world.

• Syntax vs. Semantics: Large language models process the structure of language (syntax) without grasping what those symbols actually mean in a lived, conscious reality (semantics).

• Redefining the Mind Downward: As critics observe, seeing consciousness in a chatbot does not mean the machine has risen to human sentience; it means the definition of human consciousness has been overly reduced to mere computation.

You and Hinton, are confusing the simulation of consciousness for actual consciousness.

A basic error.

4 hours ago, BritManToo said:

"The “solution” we chose was more debt."

The solution our governments chose was to fill our countries with Indians, Africans and South Americans.

The solution I would have chosen, would be to let out countries populations naturally decline and make the best of cheaper housing but a poorer standard of living. Obviously not a polular choice with out greedy masters!

Please don't include me in your 'we'.

Forest for the trees, fella.

Yes, you're upset that migrants came to the UK. They are not the cause of the world's $300 trillion debt bomb.

A couple of things have created it, not the least of which is technological change, which obviated labor to a significant degree. Speaking just for the US, one of the solutions was to create a 3rd source of household income, which was borrowing against home equity. That got a little carried away, eventually leading to an explosion of subprime loans, and then the creation of products to supposedly lessen or spread the risk, which ended abruptly in 2008.

As AI obviates more human labor---both manual and intellectual---that is going to cost society trillions of dollars. I'm not sure where that money is going to come from, but I suspect it will be freshly conjured in the same way the US has run up $40 trillion.

The qini coefficient of society is going to explode, as there will be a small number of trillionaires.....and everybody else, trying to survive off UBI. <That's the best case scenario.

12 minutes ago, Wingate said:

es, you're upset that migrants came to the UK. They are not the cause of the world's $300 trillion debt bomb.

I haven't had any debt since 1995.

A couple of things have created it, not the least of which is technological change, which obviated labor to a significant degree. Speaking just for the US, one of the solutions was to create a 3rd source of household income, which was borrowing against home equity. That got a little carried away, eventually leading to an explosion of subprime loans, and then the creation of products to supposedly lessen or spread the risk, which ended abruptly in 2008.

Again, your analysis is fantastically deficient.

The cause of the Subprime Loan crisis was NOT 3rd source household income via borrowing on home equity in response to technological change. You are just trying to fit something in a preconceived narrative, very obviously.

The real causes of the Subprime loan crisis were:

1) Ultra low interest rates made it easy to borrow and thus house prices soared

2) US banks not doing their jobs, and issuing loans to people who never should have got them

3) Banks misleading customers, and starting with low interest rates, then raising them

4) Interest rates inevitably went up again, and thus many could not afford to pay back.

You seem to develop these stories in your head and try to make the facts fit, but it's like a patchwork quilt that doesn't work.

This forum will do your head in long before a real crisis in the real world.

53 minutes ago, Cameroni said:

Don't worry, Wingate. Your electric razor won't kill you in your sleep. Nor will your Google Nest.

In a worst case scenario, if all your appliances are connected to the internet 24/7, a hacker can probably get one of your gadgets or home robot to kill you in your sleep.

Here are some of my concerns about this AI buildout.....

First, not all the AI labs nor data centers are going to win. Think of it like MySpace and Lycos. They lost out to Facebook and Google.

Second, the buildout and AI development is wildly capital intensive. Firms have gotten around some of it by making OBS commitments, as well as obtaining a kind of semi-equity/semi-debt financing from private equity (which is non-exchange traded, so its value can be whatever the private equity firm needs it to be. If you owned CDSs in 2009, you know how frustrating this can be...until the issuer hedges itself).

Most data center builders have massive long term commitments for purchasing and leasing. Some have set up external entities to obfuscate and allow their balance sheets to look much better than they are. They all do it, whether META or Oracle or Amazon or whomever. Up to now the market has just given it a wink and a nod.

META, for example, is building a massive data center in Louisiana called Hyperion. They will have a 25 year lease on it, so even if they lose the AI race, they are still legally committed. They also have purchase commitments with all the wide range of companies, foreign and domestic, who provide everything from chip racks to fiber optics to cooling systems to switches. These are legal contracts, and thus a liability, but they don't have to put it on the balance sheet. They also fiddle with depreciation, accounting for things like NVIDIA chips for much longer than the useful life. That juices the appearance of earnings. META has an entity which sure looks like a subsidiary (Blue Owl) which is where the debt to build Hyperion sits, and META has its 25 year lease with Blue Owl. (These are some of the "creative accounting" methods Japan used back in the late 1980s, before the Nikkei peaked and fell 82% over the next two decades.)

Sometimes one can find these commitments in the company's 10-K, but often they just go unstated. It takes a little digging to find out what's actually going on, and as Warren Buffett says, "You don't know who's been swimming naked until the tide goes out".

The total OBS commitments of the data center builders is around $3,000,000,000,000. Not all of them will win the AI race. Even NVIDIA has entered the fray, committing funding to OpenAI with three times the total debt NVIDIA now holds. If OpenAI fails to win the AI race...uh oh.

Now I'm sure it will all end perfectly well and NASDAQ 50,000 is on schedule for next year, but I do remember all the pros screaming that 2008's troubles would be limited to subprime and only hit 5% of loans. Countrywide, Bear Sterns and Lehman would now say otherwise.

Edited Friday at 07:51 AM4 days by Wingate

25 minutes ago, save the frogs said:

In a worst case scenario, if all your appliances are connected to the internet 24/7, a hacker can probably get one of your gadgets or home robot to kill you in your sleep.

Imagine how much work that would be for a hacker. Much easier to come to your house, break in and kill you with a hammer.

Equally AI lacks the intent to do any of these bad things, when it does do bad things, it is because of HUMAN error, mistakes where humans did not guide AI properly, formulate questions false. AI did not hack a website because it felt like it, but because of faulty human instructions.

So again, the biggest danger is stupidity. Human stupidity.

First of all, humans have not been completely dominant for 100,000 years. Our almost total dominance is relatively recent, and even then, “top of the food chain” really describes some individuals, societies and connections more than every human. And it is worth remembering that dominance can be temporary.

AI threats are also not quite like human threats. AI comes with our errors because, after all, we built it. It still needs context and objectives, and what it learns comes from human knowledge, history and the environment we give it. Where all this ends? We will see.

Most likely we will end up with AI fighting AI. Then where do humans fit into the picture? Will there be a kill switch? After all, if AI becomes truly independent, humans are basically useless to AI.

In many ways they are interlinked

8 minutes ago, Cameroni said:

So again, the biggest danger is stupidity. Human stupidity.

The HuggingFace incident - they were programmed by people. it's not AI going rogue.

But if people are too afraid of AI, they may try to block the development of data centers.

And I'm not sure what the mindset is in China. But they may get ahead in the AI race by leaps and bounds if westerners start fear-mongering about AI.

4 minutes ago, Photoguy21 said:

In many ways they are interlinked

Well, if AI ever figures out the question of what we are really fighting for, it may just decide to wipe out the real problem.

18 minutes ago, Hummin said:

Well, if AI ever figures out the question of what we are really fighting for, it may just decide to wipe out the real problem.

Which could well be humanity

6 hours ago, BritManToo said:

"The “solution” we chose was more debt."

The solution our governments chose was to fill our countries with Indians, Africans and South Americans.

The solution I would have chosen, would be to let out countries populations naturally decline and make the best of cheaper housing but a poorer standard of living. Obviously not a polular choice with out greedy masters!

Please don't include me in your 'we'.

White populations have been intentionally driven down and out by feminism, engineered housing bubbles, insane divorce laws, mass immigration (ultimately paid for by whites) etc etc etc etc -

basically a LOT of what you see is social engineering intended to stamp out the white race - allegedly.

27 minutes ago, Tourist2 said:

White populations have been intentionally driven down and out by feminism, engineered housing bubbles, insane divorce laws, mass immigration (ultimately paid for by whites) etc etc etc etc -

basically a LOT of what you see is social engineering intended to stamp out the white race - allegedly.

From what did this really evolve? From poorer and more rigid societies into much richer middle class societies with far more individual freedom. Women became less dependent on men, and relationships became more about choice. Maybe part of the problem is not some grand plan against men, but that some men never managed to adapt themselves to the new reality.

Today’s Western countries lack realistic and futuristic ideals and goals. AI represents a totally new dimension of threats, where AI against AI may eventually become the biggest problem, until they figure out why we are fighting at all and start problem solving instead.

First, not all the AI labs nor data centers are going to win. Think of it like MySpace and Lycos. They lost out to Facebook and Google.

That's a given. Just like back in the day not all railroad companies survived to be prosperous. Does that mean Railroads were a failure or caused the end of humanity? Not really.

1 hour ago, Wingate said:

Now I'm sure it will all end perfectly well and NASDAQ 50,000 is on schedule for next year, but I do remember all the pros screaming that 2008's troubles would be limited to subprime and only hit 5% of loans.

I remember you predicting total financial armaggedon a while ago, that never came to pass.

NASDAQ goes up and down, some companies fail. It is not the end of the world. We had an end of the world scenario in Germany in 1945, the country completely bombed out, and still Germany managed to come back. So that tells you enough about resilience.

Edited Friday at 09:38 AM4 days by Cameroni

1 hour ago, Hummin said:

Well, if AI ever figures out the question of what we are really fighting for, it may just decide to wipe out the real problem.

But "fighting" assumes AI has the same mentality as humans.

How do you define super-intelligence?

Maybe super-intelligence doesn't need or want to fight or kill anyone.

Maybe only stupid low-intelligence human beings (ie all of us), who are constantly in survival mode, need to constantly be in fight or flight mode and think the solution to all problems is to fight the other guy.

Edited Friday at 09:40 AM4 days by save the frogs

Impressive, we managed to get some Trump bashing in on the very first post.

24 minutes ago, save the frogs said:

But "fighting" assumes AI has the same mentality as humans.

How do you define super-intelligence?

Maybe super-intelligence doesn't need or want to fight or kill anyone.

Maybe only stupid low-intelligence human beings (ie all of us), who are constantly in survival mode, need to constantly be in fight or flight mode and think the solution to all problems is to fight the other guy.

Good question, but remember who is in charge for now. Humans are using tactics that have hardly changed since we became humans, and even before. The same tactics, the same failed results, reinvented over and over again. Now we are teaching AI those same tactics to fight an enemy.

Sooner or later AI may recognize the human error and start problem solving based on what it has learned from our history. Then what? Control humanity, or erase humanity? Humans will probably always find something to fight over, so if machines eventually see us as the pest creating the problem, what then?

What really is superintelligence? It is intelligence with far more capacity than humans, most likely able to see things objectively in a very different light and find much more precise solutions, without necessarily sharing human values or considerations.