Yves here. Rob Urie’s wide-ranging article takes on the long and (save at its outset) deliberate failure of US environmental policy and links it to the domination of the interests of capital. He then turns to AI’s underperformance for its professed uses and contends that its more plausible implementation is for mass surveillance. Urie does not tie these discussions together tidily even though he has left the informational bread crumbs to do so. The key point is one that has bizarrely been ignored in many recent debates over AI: its enormous energy use and the further climate-wrecking impact satisfying that will have (not just greenhouse gas emissions but also the need for large volumes of water for cooling, which both contaminates the water and renders it so hot that it threatens marine life when it is returned to rivers1).
This piece has some great phrase-making, like “This is why I don’t take the view that an accumulation of socialists makes socialism.” Its opening section, on the largely performative US environmental policies, dovetails with a point we have made often”: that even somewhat stronger-form proposal like the Green New Deal, are climate change hopium. It sells the idea that it’s possible to preserve modern lifestyles and avert worst outcomes by making some not-painful changes and investments. We have argued that the only approach that has a chance of being adequate in reducing greenhouse gas emissions is radical conservation. We are on a path to the dangerous fallback of widespread geoengineering. So by not getting in front of catastrophic climate changes soon and seriously enough, people all over the world will have changes, often very destructive, forced on them.
One place where I differ with Rob is the implicit claim that there is some sort of higher-order plan among the elites regarding AI development and deployment. Having had some exposure to venture capital investing (I briefly screened deals for what was then a large fund), the Dune saying, “The most persistent principles of the universe were accident and error,” applies here. The venture and tech communities desperately needed a big new set of fundraising/investment opportunities. The supposed marvel of AI fit that need.
And whether it really makes sense is not as relevant as most assume. As one private equity investor said in dismissing the message from his team, that the assumptions behind a prospective deal were flawed, “I’m selling the sizzle, not the steak.”
By Rob Urie, author of Zen Economics, artist, and musician who publishes The Journal of Belligerent Pontification on Substack
In 2017 I wrote an article arguing that neoliberal climate policies were wholly inadequate to address the scale of environmental problems and that there existed a brief window before capital reasserted itself to end the entire effort. Noam Chomsky took the other side of the argument— which was still allowed back then, claiming that an economist at UMASS-Amherst had a detailed solution to climate change that only required a few small tweaks to solve it.
The economist— Robert Pollin, was elevated to be Joe Biden’s ‘climate guy.’ Pollin’s effort yielded exactly the product that I claimed that it would in the article— absolutely nothing. But getting the direction right wasn’t my point. Understanding and publicly conveying why it was the case was the point. Pollin’s plan was ‘pragmatically’ minimalist. Do the least that could possibly address the problem and forget it. But it predictably (I predicted it) went nowhere.
The problem with Pollin’s solution is that it ignores the power of capital to derail it. I argued that capital would turn any public policy into a way to enrich America’s oligarchs, and that this would prevent it from being effective. Before Joe Biden took office I went through his environmental proposals in detail. There was no detail. The conclusion was that they weren’t serious. Their top priority was making Biden’s donors rich using public money.
Those reading this in a political party frame need to stop doing so now or seek psychiatric help. I’ve followed environmental issues since 1970. As I explain below, the first head of the EPA was a Republican. The party difference then was over how to accomplish environmental goals, not whether or not to pursue them. And the Democrats have been putting forward only Republican proposals since Bill Clinton was in office. So, the division is between when both parties cared about environmental issues and today when neither party cares about them.
Republicans stopped being environmentalists with Ronald Reagan’s election in 1980. The first head of the EPA, William Ruckelshaus, was both an environmentalist and a Republican. Environmentalism was a bi-partisan affair until Reagan mocked it in a political move to exploit the class divide between relatively well-to-do urban environmentalists and the loggers and coal miners losing their jobs from top-down environmental legislation.
As with Robert Pollin, the solutions were directionally correct while being politically counterproductive. If I wanted to create a large group of permanent political enemies, I would pass legislation intended to clean the air in New York and Los Angeles while economically devastating large swaths of the heartland. The emergence of the current urban/rural political divide began here. The arrogance of those willing to end other people’s livelihoods for what to them is an abstract benefit is breathtaking,
US greenhouse gas emissions were eventually outsourced to China. While they had been moderating with the Great Recession and the Covid-19 pandemic, US greenhouse emissions are now expected to rise 12% per year from AI. The setup is that a few dozen tech billionaires think that they aren’t rich enough yet and therefore wish to build a slight variation on existing technologies that require trillions in environmentally destructive infrastructure. Because they are billionaires, their dream of building crap that nobody asked for and that doesn’t work is becoming a reality.
This was my point in the 2017 article. Without bringing capital to heel, there is no realistic environmental path forward. Pollin’s ‘pragmatic’ approach enriched a few green tech executives. Doing so was the entire point of Biden’s adoption of Robert Pollin’s program. It looked complicated enough so that eyes would glaze over. But at its core, it was device to transfer wealth from the American people to Biden’s donors. That Pollin didn’t understand this is what made him an economics professor at UMASS – Amherst.
Donald Trump is the product of this class-war environmentalism. Displaced coal miners and loggers could have been given jobs doing other necessary things. The New Deal programs that emerged from the Great Depression built much of the infrastructure that today makes life in older American cities livable. Policy wonks don’t get Trump’s appeal because they’ve never had people three thousand miles away from them end their livelihoods, marriages, communities, and futures.
To understand the imperious mindset of doing so, consider Robert Moses, the New York oligarch who destroyed much of New York City in the 1960s by building highways through important economic routes. The resulting devastation was treated as an act of nature when it hit even though it was entirely oligarch-made. Poor people were blamed for the economic destruction that Moses was entirely to blame for. This was the class position behind early environmental legislation.
None of this is to argue against environmental policies in the present. The idiocy of ignoring accumulating environmental destruction is that the costs of remediating it are growing exponentially. Left ‘abundance’ theory needs to explain how it gets the world on a sustainable path forward. The irony is that quite a bit less would have needed to be done had Pollin’s weak tea proposals been implemented. But they were structured as gifts to donors rather than as serious environmental proposals.
Readers are invited to peruse the literature on tax incentives used to further public policies. As with AI, the literature in support of tax incentives is theoretical, and therefore prospective. The empirical evidence challenging the efficacy of tax incentives is based on outcomes. In the history that I was able to find, tax incentives went mostly to executive compensation. In American fashion, they are a cynical grift, repayment from politicians to donors for campaign contributions. This was known in 2017.
The appeal of Pollin’s argument is easy enough to understand. By limiting the problem as climate change, solutions appear easy and cheap. This is the ‘pragmatic’ counter to those of us waving our arms claiming that the power of capital must be addressed before any progress can be made. And it fits into the neoliberal framing of the Democrats’ ‘green capitalism’ as the only allowable solution to environmental woes. The problem with green capitalism is that it replaces one set of problems with another. Rather than removing the causes of environmental woes, green capitalism seeks to address them after the fact.
The starting position of capital is to solve existing problems by creating new problems. Rather than limiting the production of greenhouse gases, the green capitalist solution is to spend trillions on ‘carbon capture’ technologies and EVs. Question: how is this more environmentally efficient than directly ending greenhouse gas emissions? The answer: it isn’t. Building these new technologies while maintaining the old is wildly inefficient.
Consider the current war for control of oil in and around Iran. Then add the build out of an EV industry. Are greenhouse emissions rising or falling in this scenario? They are rising. Because China builds quality EV’s at 20% of the cost of US producers, the environmentally constructive solution would be to have the Chinese build the global EV supply. The longer-term problem is called the Jevons Paradox. Making a quality product at a low price means that more will be sold, meaning more pollution, not less.
The ultimate environmental constraints are physical. The earth will either be livable for humans or it won’t be. The available evidence suggests that what could have been solved with tweaks forty years ago will require major interruptions to Western lifestyles going forward. It isn’t the environmentalists who raised the cost of resolution here. It was those who resisted the tweaks that made the problems so much more costly to solve in the present who did so.
While this is just a rumor, I’ve been hearing it for years. Donald Trump is claimed to be of the view that climate change is real. But at eighty years of age, he just doesn’t care. He’ll be dead before mass dislocations from environmental collapse take place. Charles Darwin united humanity via ‘species.’ Donald Trump is the anti-Darwin. To Trump, the rest of us are just a collection of losers. As a metaphor for the age, it fits. Israel Funds AI Lie-Bots
Recent reports have Israel ‘spiking’ AI training data with pro-genocide talking points. Readers will recall that LLM (large language model) AI is a probability model that puts forward the most prevalent words and strings of words (tokens) found in training sets as query responses. This makes Israel gaming LLM AI so predictable as to be mundane. Israel spent just under $50 million to flood AI training data with Israeli propaganda. Missing from news reports is that pro-genocide reporting by state-adjacent media in the US had already spiked the data.
Through both prevalence in training data and a hierarchy of ‘trusted sources’ that elevate the CIA, the New York Times, the Washington Post and the Pentagon in AI query results, generative AI was already the most powerful tool for distributing state and corporate propaganda ever conceived. That a fascist state (Israel) currently engaged in ethnic genocide can buy its way to legitimacy in AI query results makes perfect sense. Ninety-two percent of AI users trust the first AI query response that they get to be true.
Compare this against only 28% of Americans who trust the mainstream press. AI is structured to draw almost exclusively from US state propaganda and US state propaganda as it is filtered through the establishment press. Only 28% of Americans trust it because it is state propaganda. And yet when AI spits out this very same state propaganda it is considered to be true by 92% of AI users. This is why AI is backed financially and politically by the CIA, the Pentagon and the intelligence-adjacent American press.
Having previously queried AI related to the US – Israeli genocide in Gaza, the Israelis are wasting their money. The query results from the Google Gemini and OpenAI ChatGPT AI models were already so lopsided towards Israel that there is no more room to move in that direction. The initial layer of lies included previously debunked claims regarding October 7, 2023. The second layer included imagined threats to Israel that are Israeli projection. Israelis are worried about being attacked by genocidal fascists— like Israelis.
That AI ‘truth’ is being gamed points to the political premises that led to probability being the engine driving AI. The necessary premise behind the model choice is that truth lies at the center of the broad social discourse as it is filtered through the internet (as AI training data). This itself is premised in the equitable distribution of political and economic power. In fact, the US has amongst the most concentrated skews (towards the rich) of political and economic power in the world. Prevalence in the discourse in this case means the voice of the rich and powerful.
This isn’t my opinion. It is the math. The probability model is used to count AI tokens. AI tokens are determined by proximity in semantic groupings. In fact, meaning (semantics) for AI is the product of this proximity. The more instances of particular tokens related to a query, the higher the likelihood that an AI model will put that forward as the answer. On matters of war and peace, prevalence in training data accrues to the New York Times, the Washington Post and the Pentagon. Behind the scenes is the CIA writing the script.
The political benefit of Chatbot AI to existing power should be obvious. Having followed US foreign policy for decades, I don’t trust one word about US foreign policy from any of the organizations listed above. Where are Iraq’s WMDs? Of what relevance is the demise of Iran’s navy when its missile defense is its strategic advantage? And if Mossad is clever, why did it come up with the dumbest plan of attack (against Iran) in modern military history? So, AI sourcing has fools lying about events that they did not anticipate as its go-to on US foreign policy.
While it initially made sense to me, and based on alleged evidence, AI coding (writing computer code) seemed promising. But the problems being encountered are illustrative. If AI can pop out code that is 97% accurate in a few seconds, is that useful? As one who coded when one additional space in five thousand lines of code could take hours to find and debug, finding the 3% (1-.97 = .03 = 3%) that is wrong can require acquainting oneself in intimate detail with the 97% that is correct. What does wrong even mean here? AI boosters claim that these are growing pains that will soon be resolved. And there is technological history that supports the claim. The early voice assistants that would answer corporate phone lines and direct calls were mercilessly mocked in the film Burn After Reading. But these days, their ability to direct us into a Kafkaesque hell when we are forced to interact with them is much improved. My own motivation to find the people who created them and pull them through the phone to let them know what I think of their handiwork is much higher now.
Imagine senior coders being shifted from writing code to ‘fixing’ AI generated code. The bosses have no framework for understanding the nature of the problem. Having been the guy who explained slightly complex concepts to senior management, bosses are chosen for their ability to convey not-quite-true explanations of company products to gullible outsiders, not to understand how the products work. The point: there are a thousand ways that 97% correct AI generated code could take longer to fix (and here) than writing the code correctly from scratch.
Back to Israeli lie-bots for a moment. Google crapified its search engine to raise profits. It had a product that worked and it had the relationship with US intelligence agencies needed to chase away its equal to or better than Google competitors. If AI companies see other companies and nations paying good money to spike AI training data, why wouldn’t they just cut out the middleman and sell Chatbot results to the highest bidder? If the retort is the threat of AI losing its credibility, you have probably never used a chatbot. Credibility isn’t what is being sold.
Again, 92% of Americans believe the same query results when AI gives them that only 28% believe when originally ‘reported’ in the establishment press (links above). What this suggests is that users trust the news source, not the news. But with the AI probability engine in hand, it is the same news. This brings forth the saying ‘putting old wine in new bottles.’ What it also suggests is that at least 64% (92% – 28%) of AI users in the US have no idea how AI works. It took from 2003 with the Iraqi WMD deception until recently for the establishment press in the US to be completely and utterly discredited. The view here is that AI will never be trustworthy.
Despite the rollout of generative AI from China in the US, the Chinese use of automation AI represents the only demonstrated productivity boost that AI has to offer. This written, because Chinese AI developers were forced through chip sanctions and lower initial funding to be clever in developing AI products, they have a pricing advantage that begs the question of how Western AI will compete? Having tried Google’s Gemini, OpenAI’s ChatGPT and Anthropic’s Claude models, the output of all three is not-ready-for-prime-time.
The level of prevarication amongst the major models (ChatGPT and Claude) is Trumpian. Obviously trained to manage questions about AI gingerly (grudgingly), the rhetorical tendency to endlessly parse questions to convince the user that it is right and the user is wrong conflates the method of brute force computing with truth generation. To give credit where credit is due, brute force computing produces occasionally interesting results. But I can find better occasionally interesting results from walking in the woods than by wasting time with chatbots.
What adjustments could be made to assure that generative AI is more than just a state propaganda distribution device? Well, the AI industry currently has thousands of engineers working to solve the problem. The prior sentence is a joke. From my interactions with (Google Gemini) AI, there is a plan to form a committee to look into the matter. In fact, when the free money dries up, count on US based AI companies to be selling prime placement in AI chatbot results to anyone with the money to buy it. Even genocidal fascists, apparently.
An obvious development will be chatbot query management tools that promise for a fee to get specific chatbot results to the front of the line. The result will resemble the devolution of Google’s search algorithm from its 1997 mediocrity to the worse-than-useless advertisement delivery device of its current search engine. Perhaps Israel can be grouped with the KKK and the American Nazi Party on the client roster of some rogue DC law firm as underappreciated institutions that only need an AI advocate to make them respectable.
Having recently tested (done battle with) OpenAI’s ChatGPT and to a lesser extent Anthropic’s Claude AI models, there is no visible path that I see to chatbots producing reliable results. The training set problem is GIGO (garbage in, garbage out). Who imagined that trawling the entirety of the internet would yield anything resembling truth? Chatbots can find a recipe for chicken cacciatore in one-hundredth of a second. But the original Google search function could do so more reliably in two seconds. Is the difference in computing time really material?
Consider the Israeli query result purchase. Fifty million dollars was put toward overrepresenting Israel’s pro-genocide talking points. But is this functionally different from the internet search engines repressing or boosting reporting based on the political goals of the Federal agencies that get the internet companies to boost or repress certain views? This creates the pool of censored, boosted or repressed stories that will be represented by volume in future query results. It adds an entire layer of untrustworthiness to chatbot query results.
There are legitimate uses of AI. Process automation in manufacturing is one of them. Answering questions about the US – Israeli war against Iran with Pentagon boilerplate isn’t. As one who was raised on the New York Times and the national magazines, I read them today to understand what the CIA and Pentagon are putting forward as propaganda, not for information about US foreign policy directly. The distinction is crucial. The prior involves misdirection while the latter is an effort to explain.
Americans should prepare themselves for the AI bet to head south in two ways. The first is that the technology has a much smaller constituency than current hype suggests. Reports of productivity gains from AI are weaker and more qualified than boosters are suggesting. The second fault line is AI financing. Too much money is chasing too few opportunities. With little evidence to date that generative AI really is a product that can be sold at a profit, there is no path forward visible to yours truly that doesn’t involve a massive revaluation of AI downward.
Centrists Have Eviscerated the Democratic Party
It is difficult to imagine what Americans would consider to be a political emergency if not the present? The US is in the process of losing two wars of choice entered into by representatives of both political parties. The US is four decades into all of its economic gains going to the oligarchs. In the last ten years inflation has cut the purchasing power of American savings by 30%. The last two presidents have been amongst the most reviled in history. The US Congress has a 15% approval rate and a 79% disapproval rate. And the only reason to vote is to prevent whichever party is in power from doing something life-threateningly stupid.
In this context, the self-styled ‘centrists’ who have gutted the Democratic Party (graph below) are claiming that they will bring down the nascent ‘left’ in order to protect the interests of financial oligarchs. The leaderships of both parties support the wars that the US is currently losing. The dice have been rolled and AI will either be the future of the American economy or nothing will. And despite the moronic, fascistic, chaos of Trump, it is unclear at present if the Democrats can even manage to take the House in the midterms.
Should Trump retain control of both houses of Congress through the midterms, expect more war, more ICE murders, mass arrests of Americans for alleged political crimes and for what little hasn’t been looted to date to find its way into Trump’s pocket. That self-described centrists in the Democratic party see a few primary victories by the DSA as the problem in need of a solution demonstrates everything that I’ve written about them over the last fifteen years. Why ‘the left’ ever saw Democrats as fellow travelers is a mystery.
Graph: Democrats gained electoral ground in 2004 – 2005 as George W. Bush’s botched war against Iraq led to widespread disillusion with Republican policies. Then in 2008 Barack Obama was elected to fix the GFC (Global Financial Crisis) / Great Recession. After bailing out Wall Street and quickly pivoting to austerity for the rest of us, Obama oversaw the greatest exodus of Democrats in modern history. This is why the Democrats pivoted to identity politics after 2016. They needed to explain Obama’s economic policies in non-economic terms to ever win another election. Source: gallup.com.
The current move by centrists to force imagined lefties out of the Democratic party follows its successful removal of Bernie Sanders as the voter’s choice to represent the Democrats in the contest for the presidency in both 2016 and 2020. Regarding the 2020 election, I explained why ‘the electable candidate’ (Biden) would be the end of the Democrats in multiple missives leading up to the 2020 election. And the Democrats would be far ahead today if Bernie had burned the party to the ground following the 2020 effort that removed him.
Given the centrists’ record of eviscerating the Democratic party (graphs above, below), one would imagine that they represent the fringe views of a few dozen tech and financial oligarchs. And they do. The only thing that centrists are the center of is the far right of the party. Should this read as descriptively inaccurate, ‘electable centrist’ Joe Biden co-launched an ethnic genocide in Gaza as he undermined his own policy proposals out of worry that they would exacerbate the Federal budget deficit. Austerity to fund genocide equals fascism.
Graph: voters have been fleeing both of the duopoly parties to register as ‘Independents’ since 2008. The trend started when Republicans started fleeing the Republican party in 2004 – 2005 over the botched US war on Iraq. Voters started fleeing the Democrats 2009 – present after Barack Obama bailed out Wall Street but then quickly pivoted to austerity for the rest of us. With no Independent party to field candidates, American voters are stuck with candidates chosen by the duopoly parties. Independents are now twice the size of the duopoly parties. Source: gallup.com.
The point is that Democratic party centrists aren’t the ideological center of anything. They are a fringe element representing the views of a few percent of Democratic voters. That they are treated as the voice of the Party means that they represent to the views of the oligarchs as filtered through the news media, not those of the party rank-and-file. While my suggestion for the last year has been to split the government by electing Democrats to Congress in the midterms, I am fine voting for a presidential candidate who will lose in 2028 rather than centrists whose politics I do not support.
That centrists still have a voice after having elected Donald Trump to the White House twice is remarkable. What does this mean? Well, why are so many voters fleeing the Democratic (and Republican) party? Wouldn’t actual centrists be drawing voters into the party because it represents majority views? In fact, the self-described centrists are actively driving voters away (graph above). While I couldn’t care less about the future of either of these parties, mine apparently isn’t a fringe view. Affiliation with the duopoly parties is crashing while registered Independents are absorbing the losses. This is the definition of being a fringe party.
While I regard the efforts of a few of the newer leftie Democrats quite highly, winning primary elections points to where the counter-forces will concentrate their efforts in November. To the point made by Krystal Ball, even the eventual winners will enter Congress as junior members. Their ability to govern will depend on their ability to garner campaign contributions from people who expect political favors in return. Like the LAPD and racist cops, the system is set up to produce outcomes that few mere mortals can overcome. This is why I don’t take the view that an accumulation of socialists makes socialism. In this regard, Zohran Mamdani’s policy proposals need to be carefully watched. As one who lived in New York for decades, landlords have been out of control since the early 1980s. I have acquaintances who have paid their rent but who haven’t had heat in the winter for fifteen years. And wholesale grocery pricing favors large chains while punishing the mom-and-pop stores that serve much of the city. City-owned stores could get bulk pricing while serving underserved corners of the city.
The thing about Mamdani is that well-considered policies that solve social problems take on a life of their own. The contest isn’t right versus left or Democrat versus Republican. It is over how to make life better for Americans and the people of the world. But this isn’t how the American political system is set up. The American political system is set up to bleed us, to set us up as marks for commercial interests as self-interested politicians in the pockets of the oligarchs compel our acquiescence. America in 2026 is everything that we were warned could go wrong when we were younger.
It isn’t clear yet that Mamdani can get and keep the policies in place that will make a positive difference. His Pied-a-Terre tax for unoccupied second homes of the rich is meant to free up housing while raising revenue for the city. City owned grocery stores can raise the quality of food while lowering prices in the least served neighborhoods. And the very idea that government can solve problems rather than causing them is why the American right hated FDR and the New Deal for most of the last century.
That self-described centrists see Mamdani, rather than rule by capital, as the problem that sent the US off of the rails would be a mystery if it were a mystery. ‘Prince’ Andrew Cuomo was supposed to anointed Mayor of New York. Lo and behold, his support all came from oligarchs. When the rank-and-file had a say, that say was ‘hell no!’ The same was true with the centrist candidates in the 2024 election. They aren’t the center of anything. They are fringe representatives of the oligarchs.
If I lived in an area with an Abdul El-Sayed or Zohran Mamdani running for office, I would support them. These are the centrist candidates. They support the policies that people favor. The oligarch loving hacks who have driven the Democratic party nearly to extinction support the policies that Trump is enacting. All of the Democratic centrists planning a run for the presidency in 2028 support everything about the Trump agenda except for Trump. They oppose his vulgarity, not his policies. The choice is socialism or barbarism, not Pete Buttigieg or JD Vance.
1 Presumably the water could both be treated and allowed to cool, but the arguments made by data centers opponents suggests that neither is part of data center plans.