Nothing stops Sanders from legislating creation of a sovereign fund invested in AI labs. It can make money or lose money, and the gains can be distributed as legislated. Specifically, it would be wrong to distribute the gains to data center NIMBY's. These regions took a decision and it's paramount that they bear the consequences of their decisions. It's perfectly alright to tax at different levels and rates; let's not do that specific to AI labs. Finally, it would be honest to finally let go of the word, "creative class". Perhaps there was little creativity there to begin with.
I kinda think you conflate some thing or over-assign the AI criticism as “bubble talk”.
I think there are legitimate economically sound reasons to think it’s a bubble. It’s also true that semi-rational climate carers may use that or over-index on that as truth to avoid the tricky dichotomy you mentioned.
But I think, overwhelmingly, what’s happening is that lefty types hate AI for a bunch of unrelated reasons (climate being one used both legitimately and of convenience) and motivated reasoning their way not to “therefore bubble” but to their first principals that AI is useless in the first place.
Just a response to the first bit, that all the people who have problems with power consumption or water consumption or data center taxes are irrational: I think most such people, while plainly wrong on the object level issue of water or power or taxes, are driven by a higher level goal of "I don't like the AI world we're building, and want to stop it with any means available to me". I think intellectual dishonesty in the small undermines this goal in the large, but I at least see where they're coming from. I think if it were still 2018 and these were mostly used steaming video and syncing Excel documents people wouldn't care, but it's not, so they're fighting it where they can and how they can. This happens to be "locally, and with appeals to any perceived populist harm that can be found". Seems like a doomed strategy, but it's hard not to be at least a bit sympathetic.
AI development appears to have positive features, unequally distributed. Carbon related environmental harms (polar bears, etc.) are distributed much more broadly. It’s not obvious to me why those most impacted by harms should concede the field to those who benefit the most. Pleading that AI will result in a better future for all at some uncertain future date provides scant comfort to those, who like the polar bears, are assigned to the discard pile.
Polar bears are present even outside the arctic ice cap. I'm skeptical that they won't survive if the arctic ice cap melts down entirely. They might need some assistance from environmentalists to transport a big enough population to land within the arctic circle. The common argument is that they will lose their food source - seals, if the arctic ice cap shrinks. That may be true but seals also need the ice to give birth and rest so I assume that they'll also adapt by moving close to land ice.
Google AI alleges that in the 2024 NCAA Women's Basketball Tournament, the South Carolina Gamecocks defeated the Oregon State Beavers in the Elite Eight to advance to the Final Four.
I figured it would have happened in some other sport, but I specifically am dreaming for football, so I can also dream of commentators having the option to say things like "And the Cocks are penetrating deep into Beaver territory!" and "Oh, but in the red zone, the Beavers are now wearing the Cocks down!".
Green energy, absent an acknowledgement that there is a negative effect on grid manageability of incremental intermittent sources (solar and wind), weakens the grid and should be avoided regardless of the political justification.
"acknwledgement?" I guess yo just mean that the subsidy for non-CO2 emitting electricity generation should take accont of time of day/season, etc. as more zero-CO2 generatin at pead zero CO2 production does not reduce CO2 emissions, the object of the subsidy.
Per latest news about LLM, the Chinese yet again have come out with cheaper models (require less electricity) that are mostly just as good. And businesses are realizing that paying for the most advanced and expensive models that are only marginally better isn't worth it.
Sanders' proposal makes a lot of sense to me. Consider:
--Whether through public ownership shares or direct taxes, what makes sense is to address the huge wealth generating potential of AI and data centers with mechanisms to share that wealth broadly and to use it for good purposes.
--It also makes sense to do so before the beneficiaries of that wealth (hugely capital intensive in its origins and hugely concentrated when it flows to the owners of capital) gain its benefits and deploy their wealth to keep it in their own hands. The history of the Alaska Permanent Fund is instructive here -- if the state had not acted to enact its share of the revenues from tapping oil on state lands before that oil started flowing, it never would have come to pass.
Matt's "oh wait" objection -- we need to make sure that the revenues from these investments flow world-wide! -- is a reach, to say the least. Apply that standard to any or all of our business tax revenues and, practically, we get endless debate and no tax revenues. Let's take on this issue when the iron of public opinion is hot, get a profit-sharing revenue stream in place and then use the policy-making system that Matt knows so well to tackle the distribution of gains in future budgets.
We do not want any "business" taxes. [Property taxes as a rough fee for services rendered is OK.] We want to tax consuption of those incomes progressively.
Fending off political influence by high income people is a general problem but at least a progressive consumption tax woud mean they are taxed at a fairly high rate on money spent on exerting that influence, as that clearly would be a form of non-saving/investment = consumption).
How do we know that data centers increase emissions? Data centers may create emissions--but perhaps less than the alternative.
For many professionals, the post-Covid acceptance of Zoom/Teams has reduced the number of business trips per year. One fewer cross-country flight more than offsets a substantial amount of data center usage.
Suppose that DeepMind's material science engine discovers ways to make solar panels more efficient or cheaper. That would encourage further deployment of green energy, reducing overall emissions.
To be sure, data center usage may increase emissions, but the necessary analysis isn't provided here.
Do you think - amongst general opposition to carbon taxes generally - this kind of idea comes at an unfortunate moment where (despite massive and growing deficits) populists on both sides are racing to reduce taxes as much as possible?
Came here to say that polar bears and climate change are almost orthogonal issues, compared to the tong-in-cheek way they are portrayed here and in the general discourse.
The main problem with climate change are floods, heat waves, droughts, hurricanes, wildfires, sea level rise and other hazards that affect the economy and human beings by destroying buildings, crops, infrastructure, and directly killing or hurting people, reducing their ability to work productively, and making them sick. Not primarily about the pandas, the bears or anything.
The main threats to biodiversity are habitat loss (mainly due to agriculture, and withing that mainly due to cattle feed and pastures), overexploitation (like hunting or fishing to extinction), and only 3rd place climate change and then pollution and invasive species.
Agriculture, not energy use or data centers, is the one issue common to biodiversity and climate change. We take too much land, deforest, and replace rich habitats with monocultures of soy to feed the beef.
I love steak by the way, not here to blame people. Just saying if one likes public policy, one can analyze these things. There is a ton of noise, but also a lot of good information around.
Polar bears specifically are threatened primarily by climate change, which has directly altered the polar summer ice flows, making it more difficult for them to catch food, which is mostly marine animals, leading to starvation and weight loss, fewer births, etc.
It's true that habitat loss by human/agricultural encroachment has a large impact biodiversity, but even then, climate change often exacerbates the, for example more intense dry seasons lead animals to take greater risks to find water, often resulting in human wildlife conflict events that get them killed. They're not all independent issues.
In terms of messaging, should advocates actually call this a carbon tax, or instead some other "fee", which they have secretly calculated to equal the global social cost of carbon?
While I agree that populist backlash against data centers and AI presents an opportunity to trojan horse climate policy, I think the idea misses the forest for the trees. The looming crises that AI will bring completely overshadow its climate impact.
I found the polar bear bit very unconvincing, along with the rest of the piece really, despite my general agreement with MYs perspective on climate change and energy abundance. Sure, there are people for which preservation of species diversity or whatever is a major cost of climate change. There are also 1000 other costs not mentioned along with humanist perspective MY emphasizes, and all of us are going to assign slightly different values to those costs.
Of course, policy can't model this exactly but making an attempt at estimating it and taxing emissions accordingly, rather than aiming for a targets, is a good approach. But I'm also very skeptical of federal legislation which effectively does so narrowly on one industry without a ton of loss, weird exceptions/gaming the system, and so on. Not to mention, in the years it would take for congress to get its shit together on something like this, I would expect the extreme anti AI/data center fervor of the moment to cool down and make this whole approach silly. Or 30% will have lost our jobs, in which case this would be way to narrow a solution anyways.
I disagree that the optimal amount of warming has gone up. We actually could have global cooling if we wanted.
In 1936, when Italy started allying with Germany, FDR did not conclude that "the optimal level of global fascism has now gone up." It's true that storing our carbon emissions is technically difficult and expensive, but then again, so was fighting fascism.
?? He's just saying that the trade-off between benefits from general growth in income vs benefits from reducing CO2 emissions shifts slightly toward growth.
And I am saying that there doesn't have to be a tradeoff. I grant there is one now. But if we had excess energy from solar and nuclear, we could have all the AI the market could bear, with zero warming. Or if we had a technology to mop up all the excess anthropogenic CO2 and then some, which is technically feasible, we could have zero warming. The tradeoff is there because we won't get rid of the tradeoff.
Nothing stops Sanders from legislating creation of a sovereign fund invested in AI labs. It can make money or lose money, and the gains can be distributed as legislated. Specifically, it would be wrong to distribute the gains to data center NIMBY's. These regions took a decision and it's paramount that they bear the consequences of their decisions. It's perfectly alright to tax at different levels and rates; let's not do that specific to AI labs. Finally, it would be honest to finally let go of the word, "creative class". Perhaps there was little creativity there to begin with.
I kinda think you conflate some thing or over-assign the AI criticism as “bubble talk”.
I think there are legitimate economically sound reasons to think it’s a bubble. It’s also true that semi-rational climate carers may use that or over-index on that as truth to avoid the tricky dichotomy you mentioned.
But I think, overwhelmingly, what’s happening is that lefty types hate AI for a bunch of unrelated reasons (climate being one used both legitimately and of convenience) and motivated reasoning their way not to “therefore bubble” but to their first principals that AI is useless in the first place.
Just a response to the first bit, that all the people who have problems with power consumption or water consumption or data center taxes are irrational: I think most such people, while plainly wrong on the object level issue of water or power or taxes, are driven by a higher level goal of "I don't like the AI world we're building, and want to stop it with any means available to me". I think intellectual dishonesty in the small undermines this goal in the large, but I at least see where they're coming from. I think if it were still 2018 and these were mostly used steaming video and syncing Excel documents people wouldn't care, but it's not, so they're fighting it where they can and how they can. This happens to be "locally, and with appeals to any perceived populist harm that can be found". Seems like a doomed strategy, but it's hard not to be at least a bit sympathetic.
AI development appears to have positive features, unequally distributed. Carbon related environmental harms (polar bears, etc.) are distributed much more broadly. It’s not obvious to me why those most impacted by harms should concede the field to those who benefit the most. Pleading that AI will result in a better future for all at some uncertain future date provides scant comfort to those, who like the polar bears, are assigned to the discard pile.
Polar bears are present even outside the arctic ice cap. I'm skeptical that they won't survive if the arctic ice cap melts down entirely. They might need some assistance from environmentalists to transport a big enough population to land within the arctic circle. The common argument is that they will lose their food source - seals, if the arctic ice cap shrinks. That may be true but seals also need the ice to give birth and rest so I assume that they'll also adapt by moving close to land ice.
As a loyal member of the baby seals club of the Arctic Circle, we respectfully say “screw the damn polar bears!”
Seen on a t-shirt: “Save a tree; eat a beaver”
Good, but this will always be the best beaver joke: https://www.youtube.com/watch?v=AvWfbIe4X_4
Unless I finally get my college football dream matchup of Oregon State vs. South Carolina.
Google AI alleges that in the 2024 NCAA Women's Basketball Tournament, the South Carolina Gamecocks defeated the Oregon State Beavers in the Elite Eight to advance to the Final Four.
I figured it would have happened in some other sport, but I specifically am dreaming for football, so I can also dream of commentators having the option to say things like "And the Cocks are penetrating deep into Beaver territory!" and "Oh, but in the red zone, the Beavers are now wearing the Cocks down!".
Since your reply isn't hidden, we can determine that the Substack automated moderator is not very attuned to double entendres.
I first read "baby seals club" as "club baby seals", which was the exact opposite of what you're advocating...
Seal hunters are not brave and bold,
They murder her poor wee seals
And cut off their skins to be sold,
Cursed be he who deals.
There's no reason for this slaughtering,
They're left on the rocks to bleed
He's not a man who does this thing,
It's a cruel and a heartless deed,
A cruel and a-heartless deed.
-- from "Celia of the Seals," by Donovan
Green energy, absent an acknowledgement that there is a negative effect on grid manageability of incremental intermittent sources (solar and wind), weakens the grid and should be avoided regardless of the political justification.
"acknwledgement?" I guess yo just mean that the subsidy for non-CO2 emitting electricity generation should take accont of time of day/season, etc. as more zero-CO2 generatin at pead zero CO2 production does not reduce CO2 emissions, the object of the subsidy.
Thank you for your interest in the planet.
Per latest news about LLM, the Chinese yet again have come out with cheaper models (require less electricity) that are mostly just as good. And businesses are realizing that paying for the most advanced and expensive models that are only marginally better isn't worth it.
Sanders' proposal makes a lot of sense to me. Consider:
--Whether through public ownership shares or direct taxes, what makes sense is to address the huge wealth generating potential of AI and data centers with mechanisms to share that wealth broadly and to use it for good purposes.
--It also makes sense to do so before the beneficiaries of that wealth (hugely capital intensive in its origins and hugely concentrated when it flows to the owners of capital) gain its benefits and deploy their wealth to keep it in their own hands. The history of the Alaska Permanent Fund is instructive here -- if the state had not acted to enact its share of the revenues from tapping oil on state lands before that oil started flowing, it never would have come to pass.
Matt's "oh wait" objection -- we need to make sure that the revenues from these investments flow world-wide! -- is a reach, to say the least. Apply that standard to any or all of our business tax revenues and, practically, we get endless debate and no tax revenues. Let's take on this issue when the iron of public opinion is hot, get a profit-sharing revenue stream in place and then use the policy-making system that Matt knows so well to tackle the distribution of gains in future budgets.
We do not want any "business" taxes. [Property taxes as a rough fee for services rendered is OK.] We want to tax consuption of those incomes progressively.
Fending off political influence by high income people is a general problem but at least a progressive consumption tax woud mean they are taxed at a fairly high rate on money spent on exerting that influence, as that clearly would be a form of non-saving/investment = consumption).
How do we know that data centers increase emissions? Data centers may create emissions--but perhaps less than the alternative.
For many professionals, the post-Covid acceptance of Zoom/Teams has reduced the number of business trips per year. One fewer cross-country flight more than offsets a substantial amount of data center usage.
Suppose that DeepMind's material science engine discovers ways to make solar panels more efficient or cheaper. That would encourage further deployment of green energy, reducing overall emissions.
To be sure, data center usage may increase emissions, but the necessary analysis isn't provided here.
Do you think - amongst general opposition to carbon taxes generally - this kind of idea comes at an unfortunate moment where (despite massive and growing deficits) populists on both sides are racing to reduce taxes as much as possible?
Came here to say that polar bears and climate change are almost orthogonal issues, compared to the tong-in-cheek way they are portrayed here and in the general discourse.
The main problem with climate change are floods, heat waves, droughts, hurricanes, wildfires, sea level rise and other hazards that affect the economy and human beings by destroying buildings, crops, infrastructure, and directly killing or hurting people, reducing their ability to work productively, and making them sick. Not primarily about the pandas, the bears or anything.
The main threats to biodiversity are habitat loss (mainly due to agriculture, and withing that mainly due to cattle feed and pastures), overexploitation (like hunting or fishing to extinction), and only 3rd place climate change and then pollution and invasive species.
Agriculture, not energy use or data centers, is the one issue common to biodiversity and climate change. We take too much land, deforest, and replace rich habitats with monocultures of soy to feed the beef.
I love steak by the way, not here to blame people. Just saying if one likes public policy, one can analyze these things. There is a ton of noise, but also a lot of good information around.
Polar bears specifically are threatened primarily by climate change, which has directly altered the polar summer ice flows, making it more difficult for them to catch food, which is mostly marine animals, leading to starvation and weight loss, fewer births, etc.
It's true that habitat loss by human/agricultural encroachment has a large impact biodiversity, but even then, climate change often exacerbates the, for example more intense dry seasons lead animals to take greater risks to find water, often resulting in human wildlife conflict events that get them killed. They're not all independent issues.
In terms of messaging, should advocates actually call this a carbon tax, or instead some other "fee", which they have secretly calculated to equal the global social cost of carbon?
While I agree that populist backlash against data centers and AI presents an opportunity to trojan horse climate policy, I think the idea misses the forest for the trees. The looming crises that AI will bring completely overshadow its climate impact.
Or benefits :)
I found the polar bear bit very unconvincing, along with the rest of the piece really, despite my general agreement with MYs perspective on climate change and energy abundance. Sure, there are people for which preservation of species diversity or whatever is a major cost of climate change. There are also 1000 other costs not mentioned along with humanist perspective MY emphasizes, and all of us are going to assign slightly different values to those costs.
Of course, policy can't model this exactly but making an attempt at estimating it and taxing emissions accordingly, rather than aiming for a targets, is a good approach. But I'm also very skeptical of federal legislation which effectively does so narrowly on one industry without a ton of loss, weird exceptions/gaming the system, and so on. Not to mention, in the years it would take for congress to get its shit together on something like this, I would expect the extreme anti AI/data center fervor of the moment to cool down and make this whole approach silly. Or 30% will have lost our jobs, in which case this would be way to narrow a solution anyways.
I disagree that the optimal amount of warming has gone up. We actually could have global cooling if we wanted.
In 1936, when Italy started allying with Germany, FDR did not conclude that "the optimal level of global fascism has now gone up." It's true that storing our carbon emissions is technically difficult and expensive, but then again, so was fighting fascism.
?? He's just saying that the trade-off between benefits from general growth in income vs benefits from reducing CO2 emissions shifts slightly toward growth.
And I am saying that there doesn't have to be a tradeoff. I grant there is one now. But if we had excess energy from solar and nuclear, we could have all the AI the market could bear, with zero warming. Or if we had a technology to mop up all the excess anthropogenic CO2 and then some, which is technically feasible, we could have zero warming. The tradeoff is there because we won't get rid of the tradeoff.