BlackRock's Larry Fink urges trillions in AI investment to keep US ahead, citing national security risks if China advances. His proposal to use retirement savings sparks social media uproar.
As an aside, these types love to see ML development as a money problem: the more you throw at it, the more results you get.
That’s not how it’s been working in practice, though.
More resource constrained companies have gotten thrifty or creative. They’re forced to be efficient and realistic, and to look at interesting papers to fold into their stuff.
At the other extreme, there’s mounting anecdotes that Meta, XAI, and maybe others are spending many man hours simply trying to scale across huge clusters, or even giving GPUs “busywork” to meet resource utilization contracts. That they’re using old architectures. They’re using orders of magnitude more resources to make just slightly better models than small, cheap ones. And even if that’s overblown, their leaders are making terrible decisions, with nutty machinations, and aren’t getting punished for it because they have too much money.
It’s also fueling literal sociopaths like Altman as they do their best to snuff out competition.
All this capital is poisoning the AI space in the US. And it’s going to the wrong places.
Regardless of how everyone feels about “AI,” I believe most of the US is going to fall behind because of this. Except maybe Google.
Counterpoint, America has essentially brute-forced its way into every major technical advancement it has made in the past 100 years.
Last game-breaking invention that didn’t adhere to this was probably out of the industrial revolution.
Atomic bomb? Throw some money at it.
Computers/Transistors? Throw some money at it.
Space race (and everything from it)? Throw some money at it.
Everything to ever come out of Bell Labs? Throw some money at it.
Honestly Bell is a unique case in industry monopolies. They put tons into R&D despite having practically 0 competition.
Point is, it’s worked well for us so far, and nobody seems to realize that the bulk of America’s success in the past 100 years was actually due to pure blind luck because the entire rest of the northern hemisphere was rebuilding from war. Why change now?
Remember when transistors were invented, and the rest of the world shrugged and said “we lost the race” and then never invented them and still plow fields by horse and candlelight?
Of course not.
Framing this as a race introduces the idea that if you don’t win you’re fucked. As if time stops. It’s a bad metaphor and it only serves to undermine rational thinking. It’s designed to convince people they need to sacrifice something. Datacenters spiked your electricity bill? That’s ok because we have to win.
The space race is pretty similar. Also worth pointing out that by every metric besides man foot+moon, the USSR won. First suborbital flight? Orbital? Satellite? Animal in space? Man in space? Moon flyby? Moon landing? USSR. Did the USA quit? Did it matter?
You can throw money at whatever you want. Framing it as a race is nonsense designed to convince people to work against thier own interests so someone else can get rich.
I think the “race” part of it is somewhat necessary as long as we have a world that revolves around a handful of economic powerhouse nation-states.
It’s a conditional requirement of capitalism. If we were unified in working towards the advancement of humanity as a whole, money be damned…sure, that’d be great. Got a little work to do before we’re there, though. In the meantime, we’re competing towards pursuit of the all-mighty dollar. And yeah, that will always mean that the individual loses out, because dollars gotta come from somewhere.
But as long as we have a capitalistic society, we are always going to be at odds with the competition, i.e. China, which means no cooperation.
Furthermore, they seem to be way better at not giving a fuck about stealing IP than we are…if we make something, they clone it and make it cheap, but that never, ever happens in reverse.
Partly because they are a bit better equipped at this point for obtaining raw materials (to put it lightly) and they have a much less expensive labor force (to put it lightly).
China is where it is very explicitly because it spent a lot of time not trying to win any race. They’re still not trying to win it.
It’s massively more cost effective to fast follow in the wake of whomever is breaking new ground. Orders of magnitude.
China has been strategically losing races. Over and over. They understand there is no value in getting there first if you can be there 6 months later cheaper.
The race is made up. It’s an excuse to fuck YOU. YOU.
As an aside, these types love to see ML development as a money problem: the more you throw at it, the more results you get.
That’s not how it’s been working in practice, though.
More resource constrained companies have gotten thrifty or creative. They’re forced to be efficient and realistic, and to look at interesting papers to fold into their stuff.
At the other extreme, there’s mounting anecdotes that Meta, XAI, and maybe others are spending many man hours simply trying to scale across huge clusters, or even giving GPUs “busywork” to meet resource utilization contracts. That they’re using old architectures. They’re using orders of magnitude more resources to make just slightly better models than small, cheap ones. And even if that’s overblown, their leaders are making terrible decisions, with nutty machinations, and aren’t getting punished for it because they have too much money.
It’s also fueling literal sociopaths like Altman as they do their best to snuff out competition.
All this capital is poisoning the AI space in the US. And it’s going to the wrong places.
Regardless of how everyone feels about “AI,” I believe most of the US is going to fall behind because of this. Except maybe Google.
Counterpoint, America has essentially brute-forced its way into every major technical advancement it has made in the past 100 years.
Last game-breaking invention that didn’t adhere to this was probably out of the industrial revolution.
Atomic bomb? Throw some money at it.
Computers/Transistors? Throw some money at it.
Space race (and everything from it)? Throw some money at it.
Everything to ever come out of Bell Labs? Throw some money at it.
Honestly Bell is a unique case in industry monopolies. They put tons into R&D despite having practically 0 competition.
Point is, it’s worked well for us so far, and nobody seems to realize that the bulk of America’s success in the past 100 years was actually due to pure blind luck because the entire rest of the northern hemisphere was rebuilding from war. Why change now?
Remember when transistors were invented, and the rest of the world shrugged and said “we lost the race” and then never invented them and still plow fields by horse and candlelight?
Of course not.
Framing this as a race introduces the idea that if you don’t win you’re fucked. As if time stops. It’s a bad metaphor and it only serves to undermine rational thinking. It’s designed to convince people they need to sacrifice something. Datacenters spiked your electricity bill? That’s ok because we have to win.
The space race is pretty similar. Also worth pointing out that by every metric besides man foot+moon, the USSR won. First suborbital flight? Orbital? Satellite? Animal in space? Man in space? Moon flyby? Moon landing? USSR. Did the USA quit? Did it matter?
You can throw money at whatever you want. Framing it as a race is nonsense designed to convince people to work against thier own interests so someone else can get rich.
I think the “race” part of it is somewhat necessary as long as we have a world that revolves around a handful of economic powerhouse nation-states.
It’s a conditional requirement of capitalism. If we were unified in working towards the advancement of humanity as a whole, money be damned…sure, that’d be great. Got a little work to do before we’re there, though. In the meantime, we’re competing towards pursuit of the all-mighty dollar. And yeah, that will always mean that the individual loses out, because dollars gotta come from somewhere.
But as long as we have a capitalistic society, we are always going to be at odds with the competition, i.e. China, which means no cooperation.
Furthermore, they seem to be way better at not giving a fuck about stealing IP than we are…if we make something, they clone it and make it cheap, but that never, ever happens in reverse.
Partly because they are a bit better equipped at this point for obtaining raw materials (to put it lightly) and they have a much less expensive labor force (to put it lightly).
China is where it is very explicitly because it spent a lot of time not trying to win any race. They’re still not trying to win it.
It’s massively more cost effective to fast follow in the wake of whomever is breaking new ground. Orders of magnitude.
China has been strategically losing races. Over and over. They understand there is no value in getting there first if you can be there 6 months later cheaper.
The race is made up. It’s an excuse to fuck YOU. YOU.
That’s fair, but in all those cases I think money was flowing to the right engineers, and the resources they need.
That’s not what’s happening now.