Merged Artificial Intelligence

[Emily's Cat]

Yeah but now we know what positrons are and how they behave, we wouldn't be calling any kind of artificial brain positronic. Anyway, we're not going to get anywhere near there any time soon.

Okay, but you know conceptually what asimov’s positronic brain was about, right? So do a ton of other people who’ve read Asimov or watched any of the movies or tv shows.

Many many fewer know anything about positrons.

For all intents and purposes, you’re busy arguing that “no-see-ums” aren’t actually invisible and a lot of the time they’re actually gnats. Technically correct, but functionally irrelevant.

 

[Emily's Cat]

ai training takes massive amounts of data - and constellations will always lag behind terrestrial ones.

the only thing Space has going for it is Solar power. But you can have that on the ground, too.

Mmm.. I’m gonna say space also has space going for it. As in lots of room. And no weather to screw up collectors. And cold temperatures (if you block/redirect solar radiation). And very few people aside from some astronomers complaining about the giant eyesore in their town. And no (or minimal) gravity to limit configuration.

Seems like space might have a few other things going for it.

 

[Emily's Cat]

Somebody doesn't have a clue about how autonomous vehicles work.

Hint - GPS is only accurate to a meter or so at best, and the lag is several seconds - useless for driving. That's why a local AI processor is required. Tesla's Hardware 5 (A15) processor in the Cybercab has a 20 millisecond reaction time, about 10 times faster than a human.

So why have Starlink?
- Because people want internet, and when the car is driving by itself there's plenty of incentive to use it.
- To keep map data up to date.
- To do 'over the air' software upgrades.
- For communication between the car and robotaxi base.
- Because ground-based wireless networks are unreliable.

Adding that gps is really only a periodic check that recalibrates vehicle nav. The majority of guidance uses dead reckoning.

 

[Emily's Cat]

Getting a sufficiently large radiator would require an awful lot of launches. The cost of the radiator alone would make terrestrial data centres more cost-effective.

I fell like maybe you don’t have a grasp on how many launches space x is already doing, nor the payload size.

 

[The Great Zaganza]

Mmm.. I’m gonna space also have space going for it. As in lots of room. And no weather to screw up collectors. And cold temperatures (if you block/redirect solar radiation). And very few people aside from
Some astronomers complaining about the giant eyesore in their town. And no (or minimal) gravity to limit configuration.

Seems like space might have a few other things going for it.

no, space around earth is incredibly crowded already. Satellites have to maneuver out of each other's way many times a year.
And multiple countries are putting Starlink-like constellations into Orbit in the coming years.
A Kessler-Syndrome within the next 25 years is pretty much unavoidable.

 

[Emily's Cat]

Also no way to fix them if a meteorite smashes a hole in one.

Why? Do you think that once it’s up there nobody can ever get to it again?

 

[Emily's Cat]

Accumulating micrometeorite strikes which decreases efficiency until the whole thing just shuts down. Or in the case of micrometeorites damaging the radiators, overheats, which could be catastrophically worse.

As evidenced by the complete inability of current satellite to function, presumably?

 

[Emily's Cat]

It's a good idea. Yes, it'll be inefficient, extravagantly expensive and essentially useless for any purpose but money laundering outside all earthly jurisdictions, but what it won't have is a captive municipality to mooch power and water off of while pushing the bill to local residents. Whatever they do to solve those issues will have applications here on the ground and for future space endeavors.

Not that Musk actually intends to build this, he's just BSing on Xitter like he always does. It's a shame about his ◊◊◊◊◊◊ up penis, he'd have wrecked less of the western world if he didn't need to compensate for it.

You know, I don’t understand this mindset. I mean, if you want to hate musk, go ahead, I’m not going to stop you.

But musk has actually built the majority of things they’ve said they’re going to build. And much of it has been pretty significant technological advancements.

I feel like a lot of people have a visceral dislike of Musk for their association with Trump, and they end up retroactively devaluing what they’ve accomplished.

 

[Andy_Ross]

Elon Musk
@elonmusk

And electrical equipment, cooling, networking and batteries

Quote

Euwyn
@euwyn
Elon Musk says the AI boom will hit a WALL by the end of 2026.
He says the whole industry is about to learn a hard lesson in hardware.
Chip output is exploding, but electricity is nearly flat everywhere outside China.

So the two lines cross this year.

He says by the end of the year, we will be making more AI chips than we can actually turn on.
Millions of them pile up with nowhere to plug in.
In his words, the bottleneck for the next year is not chips, it is raw power.
Whoever can turn on the most chips fastest wins the race.
In the end, the AI race comes down to one thing:

Raw electricity.

link

 

[Olmstead]

Good, I was worried AI wasn't destructive enough, but needing more than all the power definitiely puts me at ease.

 

[Roger Ramjets]

Getting a sufficiently large radiator would require an awful lot of launches. The cost of the radiator alone would make terrestrial data centres more cost-effective.

Like I said, an engineering problem - not physics.

Euwyn
@euwyn

In the end, the AI race comes down to one thing:

Raw electricity.

And the money to pay for it. The waste will continue until the money dries up.

 

[dirtywick]

sounds like a financial problem, not an engineering one

 

[arthwollipot]

It's all three. It's ridiculously expensive, difficult to do, and ignores certain points of basic physics.

 

[dirtywick]

look, elon musk wouldn't say he's going to do something and collect a bunch of money and then not do it

 

[Roger Ramjets]

It's all three. It's ridiculously expensive, difficult to do, and ignores certain points of basic physics.

Private investors have the right to decide for themselves how to spend their money. Maybe it doesn't pan out, but that's how capitalism works. Would you be happier if Trump decided it was needed for national security so your taxes will pay for it?

At least a bunch of engineers will get paid to push the boundaries of technology - a kind of 'trickle down' if you will. I'd rather Musk spent his money on this than meddling in politics.

 

[Orphia Nay]

Good, I was worried AI wasn't destructive enough, but needing more than all the power definitiely puts me at ease.

Yep. AI wants ALL the RAM, aLL the eLecTriciTY, all the WatEr, but no, if multibillionaires make more money, the rest of the world can suffer.

 

[Roger Ramjets]

look, elon musk wouldn't say he's going to do something and collect a bunch of money and then not do it

You're being ironical of course. Now tell us when he did that.

 

[dirtywick]

You're being ironical of course. Now tell us when he did that.

i know you really want to talk about the guy, but there's like 2 giant threads dedicated to that stuff and you've participated in them extensively. so i'm going to pass on that

 

[arthwollipot]

Private investors have the right to decide for themselves how to spend their money. Maybe it doesn't pan out, but that's how capitalism works. Would you be happier if Trump decided it was needed for national security so your taxes will pay for it?

Wouldn't be my taxes paying for it, mate.

 

[Roger Ramjets]

Yep. AI wants ALL the RAM, aLL the eLecTriciTY, all the WatEr, but no, if multibillionaires make more money, the rest of the world can suffer.

They don't make money until the AI makes a profit. Meanwhile manufacturers and utilities are happy to sell them stuff provided they pay their bills. Regulators may not be happy though, which is one reason data center buildout is stalling. Musk is hoping to capitalize on that by providing an alternative.