Bigger than governments?... AI plans spawn mega deals

[Maxmars]

We have come to a point in our technological history where only the truly committed can invest... and now that we have businesses and organizations that dwarf the economic power of any nation, they are beginning to make a very powerful move.

From ArsTechnica: AI’s hungry maw drives massive $100B investment plan by Microsoft and BlackRock
Subtitled: Investment "goes beyond what any single company or government can finance."

From CIO.com: AI to go nuclear? Data center deals say it’s inevitable

Apparently, major players in the strategic enterprise of developing AI have aligned and begun to cooperate towards the goal of getting in on the ground floor of a future digital economy, one dominated by AI 'services.'

... Microsoft, BlackRock, Global Infrastructure Partners (GIP), and MGX announced a massive new AI investment partnership on Tuesday called the Global AI Infrastructure Investment Partnership (GAIIP). The partnership initially aims to raise $30 billion in private equity capital, which could later turn into $100 billion in total investment when including debt financing.

The group will invest in data centers and supporting power infrastructure for AI development. "The capital spending needed for AI infrastructure and the new energy to power it goes beyond what any single company or government can finance," Microsoft President Brad Smith said in a statement. 

This will include a new wrinkle in the energy production world...

AWS, Microsoft, and Google are going nuclear to build and operate mega data centers better equipped to meet the increasingly hefty demands of generative AI.

Earlier this year, AWS paid $650 million to purchase Talen Energy’s Cumulus Data Assets, a 960-megawatt nuclear-powered data center on site at Talen’s Susquehanna, Penn., nuclear plant, with additional data centers planned — pending approval by the Nuclear Regulatory Agency.  

Microsoft, Google, and Nucor, a steel manufacturer, released a request for information (RFI) about clean energy, and Baltimore-based energy company Constellation responded “to the RFI with our points on advanced nuclear being a fit at existing nuclear sites,” says a spokesperson for Constellation, one of the nation’s largest nuclear power providers.

“The data economy and Constellation’s nuclear energy go together like peanut butter and jelly,” said Joe Dominquez, Constellation’s CEO during a company conference call in May.

I don't know what difficulties will come to the surface in this dream world of harmonious corporate cooperation... but I bet this isn't going to be as easy as 'plug and play.' 

Nuclear power for AI... it boggles the mind.

[FlyingClayDisk]

Skynet is REAL!

Back in 1984, and for years afterwards, Skynet was a running joke, something which clearly would never happen...or so we thought.

Now Skynet is sitting on our doorstep, wrapped up with a pretty bow on top, soon to be powered by virtually limitless sources.  Oh sure, it's not called 'Skynet' of course, but how long will it be before we (humans) can no longer control the decisions it makes for our future?

Some say this is overly dramatic; a Skynet-like reality will never happen.

In 1949 George Orwell published the literary masterpiece 'Nineteen Eighty-Four'.   For decades people said it would never happen.  Seventy years later in 2019 (the covid years), many of the basic tenets of the novel became daily societal boundaries and reality.  Now, in 2024, forty years after the first mention of Skynet in the famous 1984 movie 'The Terminator', Artificial Intelligence sits on our doorstep.

In 1945 mankind saw the destructive capability of nuclear weapons.  Only afterwards did it dawn on mankind that these weapons had the capability to extinguish mankind from the planet.  Only then did the human race wake up and start trying to figure out ways to control the use of such weapons.  Similarly, Artificial Intelligence is upon us, the great savior, but mankind hasn't yet thought how to control it...only how to create it.  Funny how that works.  We figure out how to create things to seemingly save us from the 'great war' only to later realize these same things have the capability to wipe out mankind.  This time though, the great enemy isn't another people, but rather greed of money and political power.

Artificial Intelligence isn't like nuclear weapons, not like them at all.  Nuclear weapons aren't 'intelligent', they can't detonate themselves.  Artificial Intelligence isn't a weapon at all.  No, AI is something altogether different.  AI is about making decisions, and coming up with ideas, and some say "solutions" to difficult problems.  But here's the problem; no one knows how to stop AI once it takes hold, once it starts forming relationships to other instances of super-computer based AI across the interconnected and networked world we live in.

We can already see that we lack the electrical capacity to fully realize the true 'power' of AI, so now we're going to connect nuclear power generating capability to AI to get there.  And what happens when we realize we don't like some of the decisions being made by AI (Skynet)?  How can we as a race of beings be so blind as to not realize that it's only a matter of time before AI realizes that mankind is the real problem (just like Skynet did in the fictional Terminator franchise)?  Oh, but wait, I can make a few bucks with AI between now and then, right?

Ironically, the words from the final scene of a famous 1968 movie fit unnervingly well here...

"Gawd Damn You!  Gawd Damn You All to HELL!"

[ArMaP]

Nuclear power for AI... it boggles the mind.

Not really.

One of the reasons AI is getting so much publicity is because of the hardware it needs, so people need to buy new devices to be able to use all the next AI useless thing, while big companies will use huge data centres to store the data they want their own AI models to work with.

If any company wants their one AI system they need to train it on the kind of data they need, so they need specific solutions for their specific problems.

More general problems need much more data, so it's not surprising that all the big companies are jumping on that wagon, but they need huge power supplies to power all those huge data centres.

---

We can already see that we lack the electrical capacity to fully realize the true 'power' of AI, so now we're going to connect nuclear power generating capability to AI to get there.  And what happens when we realize we don't like some of the decisions being made by AI (Skynet)?  How can we as a race of beings be so blind as to not realize that it's only a matter of time before AI realizes that mankind is the real problem (just like Skynet did in the fictional Terminator franchise)?

One thing is AI making decision, another thing is someone implementing those decisions.

As long as we do not allow AI access to the physical world I suppose we are OK.

I said "suppose" because there will always be some idiots that will follow AI's decisions without thinking, and the size of the damage they make depends on how high in the decision chain those idiots are.

[Blaine91555]

The more data gleaned under the guise of needing it for AI, the more power the government has over us. AI is a tool and like all tools it can build or kill. A little off topic but it was my first thought.

Nuclear power is the only quick solution to all of our energy needs while we await the development of new renewable energy tech. These giant corporations have pandered to the phobic crowd that wrongly fights against nuclear power. and now they are promoting it?

[Maxmars]

The more data gleaned under the guise of needing it for AI, the more power the government has over us. AI is a tool and like all tools it can build or kill. A little off topic but it was my first thought.

Nuclear power is the only quick solution to all of our energy needs while we await the development of new renewable energy tech. These giant corporations have pandered to the phobic crowd that wrongly fights against nuclear power. and now they are promoting it?

I thought the same...  endless whining about nuclear plants as kaiju-monsters... but now that it's for AI... it's all good?

Very peculiar how quickly the jello-like surface of public perception is characterized by those purporting to "report."

I'd better never hear another tale of local people freezing to death for lack of heat in the home, while they game the world with nuclear power dedicated to the virtual.

[FlyingClayDisk]

It's not necessarily AI making a decision and then 'people' acting upon it; we've had automation for decades.  Pre-programmed events based on certain conditions.  One of the big things in business sectors lately has been eliminating 'heads' (people) through automation.  How long will it be before people have that "Ah HA!" moment and figure out they can hook AI to automated processes to eliminate people executing the seemingly 'dumb' tasks.  Just look what happened on the flight decks of Boeing 737-MAX aircraft with the MCAS systems.  Smarter than the pilot, eh???  Yeah, look how that worked out, and that wasn't even AI (just automation).  How long before they put AI behind the 'SCRAM' switch on a nuke reactor?  Or behind the 'Fire' button on a nuke.  Or...behind the yoke of a commercial aircraft (because hey, AI and automation is smarter than a human, right?)

Incidentally, automation has been sneaking into commercial aviation for a quite a long while now.  Remember when there used to be a Flight Engineer on the flight deck?  Gone now, to automation.  Remember when the yoke of the aircraft used to be physically connected to the flight control surfaces?  Gone now, most fly by wire now.  Do some of these automated systems work for the betterment of systems and man?  Absolutely they do.  But what happens when you replace the human brain with AI, and then couple these things to that same automation?  Bad ju-ju will likely abound.

It's not a matter of "If"; it's a matter of "When".  Sooner or later some company will opt to save a buck and put AI in charge of a mission critical system...and then we'll be sorry.

"No", you say?  Okay, then where are these safeguards over AI today?  AI development is proceeding at breakneck pace, so where are these controls to put boundaries on what AI will be allowed to do and control?  (crickets)  That's where.

[ArMaP]

"No", you say?  Okay, then where are these safeguards over AI today?  AI development is proceeding at breakneck pace, so where are these controls to put boundaries on what AI will be allowed to do and control?  (crickets)  That's where.

The EU is thinking about it.

EU AI Act: first regulation on artificial intelligence

[Maxmars]

It's not necessarily AI making a decision and then 'people' acting upon it; we've had automation for decades.  Pre-programmed events based on certain conditions.  One of the big things in business sectors lately has been eliminating 'heads' (people) through automation.  How long will it be before people have that "Ah HA!" moment and figure out they can hook AI to automated processes to eliminate people executing the seemingly 'dumb' tasks.  Just look what happened on the flight decks of Boeing 737-MAX aircraft with the MCAS systems.  Smarter than the pilot, eh???  Yeah, look how that worked out, and that wasn't even AI (just automation).  How long before they put AI behind the 'SCRAM' switch on a nuke reactor?  Or behind the 'Fire' button on a nuke.  Or...behind the yoke of a commercial aircraft (because hey, AI and automation is smarter than a human, right?)

Incidentally, automation has been sneaking into commercial aviation for a quite a long while now.  Remember when there used to be a Flight Engineer on the flight deck?  Gone now, to automation.  Remember when the yoke of the aircraft used to be physically connected to the flight control surfaces?  Gone now, most fly by wire now.  Do some of these automated systems work for the betterment of systems and man?  Absolutely they do.  But what happens when you replace the human brain with AI, and then couple these things to that same automation?  Bad ju-ju will likely abound.

...

Ouch!  I had already begun to find it difficult to 'trust' in commercial airline operations... this only makes it more... a matter of trust.

Knowing there is no physical pilot link to the flight control surfaces makes me very leery about that trust...  but I digress...

"AI" (such as the marketing has us calling it) is bad enough... and just because the industry is stoked at the possibilities, the accountants thrilled at the profit templates, and the marketers overjoyed about all the hyperbole and fantasy they can 'sell;' doesn't really mean that it is what they say it is, it can do what they 'postulate' it can do, and they can control an independent intellect they can't monitor. 

But at least as long as it is - in reality - a collection of coordinated algorithms... they might be barking up the wrong tree.... or at least selling us an orange while calling it an apple.

[ArMaP]

"AI" (such as the marketing has us calling it) is bad enough... and just because the industry is stoked at the possibilities, the accountants thrilled at the profit templates, and the marketers overjoyed about all the hyperbole and fantasy they can 'sell;' doesn't really mean that it is what they say it is, it can do what they 'postulate' it can do, and they can control an independent intellect they can't monitor.

The accountants are just another class of employee, it's the ones that get the profits that are thrilled.