The last elected Republican President to not have a recession start during his time in the White House was Chester A. Arthur, who left office in 1885.
There are few things as reliable as a Republican President bringing a recession. What makes Trump unique is that he appears to be trying to create one on purpose.
I think he’s going even deeper, full blown depression. What better way to buy up large swaths of land in prime areas so the world can be filled with golf resorts.
"...buy up large swaths of land in prime areas..."
What a very shallow and pissant intent, unfortunately very Trump-like. I can easily imagine that, with his single dimensional thinking, he would try something like this.
Unfortunately, we've also got the tech-bros to contend with and the simmering manure-tea stew that is Project 2025.
Prof Krugman, you need to add "Catch me now I'm Falling" by the Kinks as a musical coda next.
Wyoming senator is trying (as did Utah but was turned down by the Supreme Court). Wants all public land except Yellowstone NP. Grand Teton is prime real estate but it is publicly owned along with much of the state. Prefer it stays that way.
I think that is super important and despite being erroneous is baked into consciousness that Dems are econ failures. Carter really wanted people in the US to invest in energy independence, stay the course for a correction to sustainability, to dismantle the CIA and proxy wars by the military industrial complex and instead we got 8 yrs of demented Reagan trickle down gold toilet seats on yachts and FAA Union busting all based on the the Iran contra coup.
Republicans are better at messaging than Democrats. And what’s funny is Republicans believe they are terrible at messaging because they are so invested in their asinine policies like tax cuts for the wealthy that only harm the regular wealthy while adding trillions to the debt. So if Republicans are so terrible at messaging…then why is Trump appointing so many Republican talking heads to his administration??
"There is a seeming disconnect between, on one side, the economic news (somewhat worrying) and the political news (terrifying) and, on the other, the stock market, which seems untroubled by all of it."
Consumer confidence has nosedived since Trump took office, as though people are surprised by all of this. Way too many voters believe Republicans in general and Trump in particular help the economy. It's the biggest lie of American politics.
Unfortunately Democrats like Jason Furman and Larry Summers are to blame by exaggerating the impact of the recent transitory inflation and minimizing the impact of inflation from 2005-2008. Republicans stick to talking points much better than Democrats as the people Trump has appointed talked up the Bush economy and then blamed Jimmy Carter and Clinton when the economy imploded in 2008. Democrats like Furman complained about the recent inflation and then after it has little negative impact make a big deal blaming Biden for it!?! Why don’t you focus on 2005-2008 inflationary period that peaked at 5.6% CPI in July 2008???
Obama was a mistake. We thought putting a black guy in was radical. Not only did we betray the New Deal, our fake radical has given us 20 years of white backlash and a WWC that is so blinded they can't even see they are destroying our democracy.
Very true except that the white backlash is more against wokeness and CRT. Obama made white people feel good. He made them feel broad-minded for accepting him, when he was really so unchallenging. Woke and CRT got in white people’s faces and told them they were racists and should feel guilty for their privilege. There was much for white people to examine and realize, but getting their defenses up was not the way to get through to them.
I am a centrist. I believe in (regulated) free markets. In that way I am just like Obama. The white-grievance right, I am quoting, "is anarchist." They believe in destroying our stability and our government. The Left, especially the black Left, is conservative. We want gradual progress from a stable base society. CRT is not radical. It is an academic THEORY that says racism permeates America's institutions and is self-reinforcing. Yes. Whites wrongly thought accepting Obama was radical and ended the race problem. It will take decades to fix our delusions that have given us Trump.
We need(ed) to be led by that black Left. (Too late now, at least for now.) The white Left fetishized race (thus continuing racism) and gender minorities. They needed those groups to be homogeneous groups, and to remain "marginalized," so they could be hero-allies. Voters preferred a less devious form of white supremacy.
Obama was also hamstrung by extreme republican stonewalling after losing control of the house after his first term. Specifically, them playing chicken with US government default via the debt ceiling to "starve the beast" (and damage the economy). One thing Dems did to try to get things moving is changing the senate rules so you can't filibuster appointees. They're without that tool now.
And then there's RBG...
Obama absolutely deserves blame. Quite a bit of it. But I'm not sure we could've really predicted how he would let us down when we elected him. Downside of electing someone with such a short record, I guess. I think Kamala would have been similarly hard to predict, had she won.
Politics is about WHY voters pick who they pick. We can talk about incompetent GOP dudes all day, but their voters are convinced and we need to stop ignoring what is really motivating them. They learned back in the Tea Party days to LIE if asked about race. We have to force the subject and we have to find the argument to convince them to quit....or the rich will play this dog whistle game until there is no democracy left.
not talking about the oppostion....well in this way: Obama made it easy for them to go to hicksville and holler "the black guy wants to give out welfare to undeserving minorities." And Obama was obsessed with showing them he was neutral on race. All that centrist crap he did was for nothing. Just like Clinton they hated him because blacks liked him and he was not a bend-over dog for the rich. Kamala followed his shut up about race game and Trump beat her because he was doing white grievance politics (aka dog whistling).
Interesting comparison to the fracking boom. I would point out that there was also massive capital expenditures for the internet boom — just by different companies. Qwest Communications — and others — installed massive amounts of fiber optic cable, all of which was eventually used — but which led to their bankruptcy when the uptake wasn’t nearly as quick as they had anticipated. The fiber optic cable enabled much of the expansion in data usage through 2010 and beyond — but the companies that installed it had long since gone bankrupt.
Sure, but this assumes there's widespread adoption coming, like, ever. I have the same doubts that I do about Elon's obsession with self-driving cars: is there actually that big of a demand for it and what are the odds it ends up being a glorified "self-checkout" at the grocery store?
They are certainly trying to shove AI down everyone's throats. Every company that's heavily invested in AI touts their new AI products. I think they're desperately trying to justify their expenditures.
I see scientific uses like Alpha Fold, but do we really need to improve scammers and propagandists?
FWIW, I use AI a lot in my work (software development). While I hesitate to call it "transformative," it is certainly a technology that isn't going away, at least in my industry. Of course, one industry alone can't justify all the AI investments.
Yeah, that's where I am at. The Lexis AI I use at work (I'm a lawyer) is certainly useful, but it's so inaccurate that it's not really much more useful than a Lexis search. It can help you draft a brief, for instance, but you'll get laughed out of court if you actually submit an AI brief with no editing.
I find I resort to it to try and figure out solutions for problems after my normal methods at have failed, and I thus only use it on exactly the kinds of edge cases that it's unable to handle, I suppose because there's a lack of relevant high quality training data.
I should really flip this around and use it more for easy stuff, but my confidence in it is very low as a result of these experiences.
I was thinking along the same lines - there was quite a lot of capital expenditure during the internet boom. For another aspect, there was a lot of infrastructure bought and installed at the business user level. Servers, routers, desktop computers And laptops for everybody, etc. - which was not consumables, it had some lifespan.
Yes, the initial round of companies that failed were the ones that hade been valued on “eyeballs” or how many users they had; when they went bankrupt all of their equipment went up for sale in the secondary market and the NEXT year, the companies that had actually been profitable, with a sustainable business model, ran into trouble because the demand for their products (laptops, desktops, servers, routers, etc) declined sharply. Companies like Silicon Graphics disappeared, and companies like Cisco took decades to reach their prior valuations.
As a former emergency generator tech in NYC (all data operations have to have emergency power),I remember companies folding before their installations were even complete.
Btw, Lehman failed when BoE wouldn't countenance Barclays investment. Barclays later bought the wreckage, including a new data center with 6 2800kW generators (Lehman's 1st data center, 80s, only required 3 1000 kW). Oddly, the entire physical infrastructure cost Barckays no more than the contracts of key Lehman execs, iirc.
Do you think a similar phenomenon could happen with nuclear energy firms? The recent events drive an increase of SMV plants to match the expected increase of energy requirements for AI companies, but who end defaulting after making more than was necessary in the short run— the recent Microsoft closures of new data centers being particularly instructive here. The plants are the taken over only in the long run when alternative energy needs are required, but with the firms that built them long out— just like the fiber cables.
Gates has plans to build a small nuke (molten sodium) in western Wyoming now that the coal plant is on its last legs. Lot’s of speculation but maybe for not since EV charging station buildout (minus tesla’s) has been canceled. This project is/was suppose to save this small community.
Was this the same plan he had before trump and had negotiated a deal with Xi to run these nuke pods in China but the trump admin was like "no" and reneged on the deal with Xi?
The Professor and Mr. Chanos say the capital invested during the internet boom was bottom-up and not as intense and focused as the surge leading up to AI, which is being driven big companies taking the majority of risks in the hundreds of billions of dollars...nothing like Gates starting a company from his garage.
As a computer scientist I have always thought the AI boom is mostly a scam. If you talked to most people they probably think the Large Language Models which have gotten so much attention in the tech media are actually sentient (e.g. they have achieved general artificial intelligence). Nothing could be further from the truth. I see LLMs as really complicated probabilistic search engines which consume a huge amount of computing resources for their training and deployment. So far they have failed to meet the hype. Given the huge amount of investment being thrown at this technology, it is unclear whether the payoff will be worth the cost or even if there is a payoff.
As a behavioral neuroscientist I fully agree, LLMs are not even AI, let alone AGI, and it is infuriating that people have been duped by the tech bros who definitely know better, or should. The credulous news media have utterly failed to bring any skepticism to the completely unsubstantiated claims that AGI is just around the corner. There are plenty of experts out there who know that LLMs are not AI but they never quote or apparently even talk to them. They continue to call wrong answers "hallucinations" as if these programs have some sort of brain. The general public seems to believe that we are on the verge of SkyNet when what we really have is more like Teddy Ruxpin on ketamine.
I spent 20 years in data analytics and the issue no one ever solved was “dirty data”. All data sets have it and it makes doing data analysis difficult when you know that some of your data has errors. With AI, there is a huge dependency on feeding the system high quality data. But who decides what is high quality. If the system is fed bogus research that Ivermectin wards off COVID, the AI will think that is a valid solution.
If people who don’t know what they are talking about feed the system bad data, it will make bad decisions. In other words, there is real potential that some AI systems could just become a MAGA voter that can do faster math calculations.
Agree completely and would just add that it depends on who decides if data is "good" or "bad" and if the system can tell the difference or adapt to new information. Yes 1+1=2 but much of what is considered "good data" later becomes at least incomplete if not wholly incorrect, such as trans fats being better for you than saturated fats. Beyond that there is the entire realm of opinion which is not necessarily good or bad to say nothing about emotions and experience of the physical world through sensation and perception as well as social capacities like empathy. There are people who are incapable of empathy. Those people are often psychopaths.
We are far from AGI. Us computer types do tend to anthromorphize aspects of computing. I do find the "hallucinate" metaphor to be quite misleading. What is really troubling is we do not know why the LLM came up with a wrong answer. In a system with billions of compute noses it is pretty much impossible to trace down the caue of the problem.
Thank you for that clarification because I don't know this field but it smells like desperation and smoke and mirrors, yet there are threats of AI fueled mass layoffs coming shortly and AI insurance claim travesties that are fairly well documented, and the race to the bottom previously didn't depend on any technology working as well as it could/should have. I studied under Lester Milbrath re sustainability and AI back in the 80s and he predicted eventual collapse and decentralization - people literally going back to doing their own work, but right now idk if that's where we are headed.
Thanks George. I spent 48 years in IT and have seen fads come and go. The probabalistic neural networks underlying the LLM tech is interesting and useful in some circumstances. But these systems are very expensive to train and run and the results are meiocre at best.
I personally think that there is a serious misconception about AI. There are many different approaches in AI and not all of them involve neural network. More importantly, there is not a all-encompassing AI model. In addition, to solve a problem, one does not have to use AI. A legacy application may actually be a better solution. Of course that may get one involved in a technology religion argument.
One AI model is one particular set or subset of knowledge (for lack of better terms ) in solving a problem. A knowledge domain can be considered as one particular, discipline, expertise or specialty. So Medicine is a domain. Within that domain, it can be further categorized into many subdomains: family medicine, cardiology, psychiatry, etc. The first step to build a successful model is knowledge acquisition. You have to know what problem you are going to solve, what objects and their relationship are involved and what the methods to solve the problem are. In neural network, this means the knowledge to build a training set. For instance, Language model is for language processing. The training set will require various language patterns, topics and contexts to understand the nuances of human language in order to generate the relevant responses for a question or a command. Furthermore, you cannot use a domain model for another domain. You cannot use an English model for Chinese interpretation. Even the vision model for a robot can be different from one for autonomous vehicle (vehicle in the other lane, its speed, etc). Government provides a lot of services, each of which requires different expertise (knowledge domains, e.g. taxation) and knowledge set (tax laws, accounting, etc). Do the Muskrats understand that (many of the things they did seems to indicate they do not) or does Musk simply lie to get his tax cut, to get the government contract for a multi-years and multi-billion dollars business, or all of the above. None of the new processes of a service can be developed in a short time frame and before they are developed, he disrupts or destroys many government services, infringes on people's privacy, destroy people's career, and potentially damages the countries' security.
There are many issues or challenges with AI, technically, economically, socially and in businesses in general. Without going further, I think the current events do highlight a problem. When a technology with such potential falls into the hand of some one like Musk who believes in a terrible and simplistic ideology, who has no moral compass, empathy and compassion for any one else, that can lead to disastrous consequences.
I agree there are different approaches to AI. Neural networks are but one approach. And as I said, there are domains where AI can be very useful as a tool. LLMs, however, are dominating the marketing for AI systems. Because LLMs exhibit seemingly human like behavior when it comes to written language, many folks think they are sentient and have reasoning ability. They are in fact only as good as their training sets. They are reasonably good stochastic parrots.
I agree that an application is only as good as its training sets. Important thing is that training sets can be modified and improved.
The more sophisticated applications involve the integration of multiple domains. One particular use case is in healthcare for diagnosis. Using computer vision, it can analyze a medical image such as a X-ray and feeds the output to a LLM to correlate these findings with patient history and medical research, delivering diagnostics and potential treatment options. Other noteworthy use cases are surveillance, inventory control and quality control. I do not know at what stage these application are, but they sound promising.
One thing I also learn from past experience is that we can never under-estimate the salesman of, say, Microsoft. I have seen them successfully sold very terrible products.
I treat the current rush into AI as the infrastructure building phase. Of course that will bring its own problems and issues. Along the way, there will be more confusion and chaos but there will also be technology advances. We will see changes in computer networking due to wide bandwidth requirements, environmental impact, such as using nuclear power, and advances in computer platform, such as quantum computers. At this point, I am more concerned about its abusive uses and the human cost these will bring.
You are consistently a good read. Your experience and brains have given you remarkable and extremely valuable insights. I hope you will continue to share your take on the passing scene with the public.
Honestly I think the boomers stayed healthy longer than most predicted and kept Joanne fabrics alive….gen X is probably sewing things but they will buy everything from Amazon. I can’t believe Barnes and Noble still exists when I go to the grocery store!! WTF???
As a semi-related side note to Paul's analogy, it takes a Yooooge amount of water to frack an oil well.
There is no "typical" well, thus the gallons of water/well ranges from 1,500,000 in the Bakken to 15,800,000 in the Horn River Shale in BC, CA. Most fracking water now comes from surface sources that, increasingly, are augmented by deep groundwater as drought impacts much of our nation, particularly the big fracking states. Big Oil is relying on more and deeper wells to access ancient aquifer water to force decreasing amounts of oil out of each fracked o/g well.
While the total amount of water used for fracking is relatively small compared to that used by, say, notoriously inefficient Big Ag, (BTW, much of Big Ag's agua is used to grow corn/beans to produce low-quality ethanol to burn in our 4wd, urban SUV/Pickup Fuelhogs), we still commit an estimated 125,000,000,000 gallons of our decreasing water supplies to frack wells for fossil fuels that we burn and then dump the resultant GHGs into the atmosphere of Our Only Home. Globally, we created a record amount of GHGs in 2024 (Yay, us?) - but that's a different topic.
Estimates vary widely, but, roughly, we commit enough water to supply 5,000,000 US homes/yr to frack wells for more oil/gas to burn - and further damage our environment. And, you might ask, what happens to the water that was used for fracking? Well, dang, turns out that those billions of gallons of fracking water are toxic. Disposing of that toxic water is a another Yooooge problem that results from our Fossil Fuel Addiction. Much info online about that disposal prob.
We set ourselves up for this, though. As a society we have determined that any activity that generates financial gain is inherently good. The markets are not in denial, they are taking full advantage of the chaotic situation. Eventually they will take their profits to Aruba and leave the rest of us to enjoy the smoking ruin our economy will become. And Republicans will find a way to blame it on Democrats
This is what Mark Andreesen more or less stated in his poorly written manifesto. The billionaire class thinks growth at any cost is good. And, as I often like to point out, while Andreesen is a billionaire, he is not very smart.
I am not particularly knowledgeable about economics, but I do have pretty long experience in boom and bust cycles in support for research in machine intelligence (a more inclusive term historically than AI). And that experience, which has its correlate in the commercial world, tells me that we should expect that technological successes will lead to unreasonable expectations, and that those expectations will not be realized and the wave of support will subside. Along the way, we will also be surprised by some applications working better than we expected (such as the large language models that have taken so many by surprise starting a few years ago). But overall many will think the technology can do much more than it really can.
Also, because of the huge costs involved in training (and drawing inferences from) the huge models, there will be major efforts to find cheaper/smaller ways to get similar results, which will limit the expansion of hardware requirements.
The apparent market calm (before the storm) sounds like a toned down version of "irrational exuberance".
The increase in inflation expectations is no surprise. Sticky prices plus inflationary policies are a volatile combination. Interesting that all the layoffs - both public and private - don't appear to have influenced this - yet.
Prepare for the great collapse. Musk and agency heads are at odds. Trump has no plans just bullying. Macron stood up to him, we are disconnecting from the world. Does anyone think that is a good thing?
You assume a US citizen with U$D will just easily be able to transfer that money there for Euros and be automatically granted residency. Why would France want another free health care recipient and accompanying U$Ds?
Freud coined the word “compartmentalization” to describe how many deal with cognitive dissonance. Sounds like the players Chanos describes and when you overlay “gaslighting” ie emotional fracking, looks like 2025. voila as they say.
The nub with the fracking companies is that they funded with high yield fixed rate debt to prevent equity dilution. This meant that they couldn’t leave the oil in the ground when the prices tanked. When the price drops the more they pumped the less profitable they became. New financings came with restrictions that allowed a moratorium on production based on market conditions. This is why the “drill baby drill” crowd needs to be circumspect.
An additional challenge came from oil rich countries that had lower production costs where they could just turn on the spigot and flood the market with excess oil.
But isn't there an effective monopoly consolidation here in the US among producers and price fixing non-compete with OPEC? Read something last yr re this mvmt around pricing/competition. Idk how the financing part fits in with this
I was referring to the first two crashes that I was aware of when I was in commercial banking and corporate finance.
After the crash of 2020, I do recall that there was talk among Trump, OPEC, the Russians and the frackers to limit production to keep the prices high. Hence, the runup in prices under Biden.
So, I think that there was some sort of collusion.
That we even have to clarify for each other which of several crashes in <20 yrs is what's wrong with this picture. incompetent asshats and unbridled privatization, tax cuts and profiteering. I'm not even a masters level and could see both the first and second coming, the whole feigning surprise by gvmt leadership in 2009 was infuriating.
Trump is the first president to win because oil prices were too low in 2016 and it caused a mini recession in Rust Belt states because they make fracking equipment…and then he won in 2024 because oil prices were too high not in 2024 but in 2022!?! And so 60 Minutes had a PA diner owner that voted for Trump complaining about oil prices when high oil prices are actually good for her business!?! America is the biggest oil and gas producer by far!?!
But that’s how you know Republicans are better at messaging than Democrats because they can manipulate every issue to their advantage…although having voters treat politics like football fandom helps a lot! Although most football fans know more about their team’s history than Republicans know about the current issues.
omg, these musical codas are making me feel old! i remember sitting in the theater watching this very young David Byrne give it all - can't believe that was more than 40 years ago!
I started law school in 1981, and still remember seeing a meme in 2021 that said, "Friends don't tell friends that 1984 was 40 years ago." When I saw the meme, my reaction was, "WTF, that can't be right!" (Sadly, it was, which meant that my "study group" and I went to our 40th reunion last year.)
How funny! I was in law school in 1984 too. I started in 1982, so my 40 year reunion is this year. My daughters have been referring to the 80's as the dark ages for years, making me feel like a dinosaur when I was only in my 40's...!
Excellent and convincing connection to fracking. Pull back by Mr. Softie first warning sign. Thanks Paul for your ongoing efforts to infuse sanity into the public discourse in the US
No, inapplicable and misleading. Fracking is an extractive activity. And all such activities peter out. Oil fields deplete, gold mines become uneconomical, whaling destroyed itself, etc. But car manufacturing only got bigger with time, and no reversion to the mean.
(sigh), the comparison was not AI as a commodity to oil...the comparison was on expected return. In fracking's case people expected too high a return and there was too much money chasing too small a profit. That is how Mr. Chaos expects AI to play out.
The older ones are. The new ones are packed with computer chips. When my wife bought a new car last year, we were told that it would cost $2,000 to repair/replace the computer if it started giving trouble.
In a sense AI is extractive, being dependent on all the (digital) knowledge that exists. As I see it, there just isn't a lot of new knowledge being found.
There are scientific studies where the huge computing power really does make a difference. Large data sets allow patterns to be seen.
I think the problem is AI is being sold as the be all end all for everyone instead of the scientific institutions that it will really benefit.
I don't want AI to set up a vacation for me or plan my wedding or decide where I should go to dinner tonight. I don't want it to write my stories for me or draw pictures for me.
I would like it to do the dishes and sweep the floor.
AI is comparable to fracking or streaming services in that capital assets (chips, wells, proprietary content) depreciate fast - faster than the 20%pa used for accounting purposes, meaning that the reported profits may be completely overstated to non existent.
Industrial robots used in auto manufacturing have an annual depreciation of around 15-20%, fracking wells loose about 40-60% output in one year, AI chips depreciate due to the sheer scale of technological progress (50% pa?)
Undermining confidence in the economy is easy to do and slinks along through the grass until it springs up to bite.
Things to consider:
Half of the US is terrified that Trump/Musk are driving us off a cliff domestically and internationally. How many of us (the most educated and affluent) are going to start up or expand businesses with that going on? Half of America is slamming on the brakes and shutting their wallets to hunker down for bad times to come.
The Federal government has all kinds of tributaries flowing through the economy. When it suddenly dries up people are left high and dry. In my area the people who do forest project work for fuel reduction aren't working. We wonder if there will be firefighters when the summer wildfires begin. The county government wants to lay off a big percentage of their workers because they say money won't be filtering down from the Feds. The effects are just starting to ripple through the economy.
Jim Chanos' perspective is very illuminating, thanks for introducing us to him. The fracking bubble does fit better in many ways.
Also, wanted to share 2 quick thoughts:
1) You, Jim Chanos and someone like Mark Cuban really ought to do a joint podcast to talk about the state of the economy and markets. The 3 of you would be interesting to listen to. Not a regularly scheduled thing, just something to do whenever the 3 of you could triangulate it.
2) Some weekend I hope you can share with us about how Isaac Asimov's "Foundation" novels influenced your career choices. Sci-fi and economics may seem like unlikely companions, unless you've read Asimov. 😁
He ties it into a larger picture of the 'Rot Economy' in Silicon Valley; that the singular drive to monetize EVERYTHING in IT has steadily degraded the very utility the internet wrought. Things like Google deliberately degrading their search by focusing on increasing their revenue by pushing paid content over what you actually searched for.
If you feel that your 'google-fu' has gotten worse, that you're spending more time trying to find an actual answer to your question, that you're wading through things that aren't useful to your question...IT'S NOT YOU.
And AI isn't helping this materially because AI very very much has a GIGO problem.
And all of this is before reckoning with the fact that AI is hugely expensive, without any clear way to turn that expense into something actually useful.
10 out of the last 10 recessions started under Republican Presidents. Just saying.
The last elected Republican President to not have a recession start during his time in the White House was Chester A. Arthur, who left office in 1885.
There are few things as reliable as a Republican President bringing a recession. What makes Trump unique is that he appears to be trying to create one on purpose.
I think he’s going even deeper, full blown depression. What better way to buy up large swaths of land in prime areas so the world can be filled with golf resorts.
"...buy up large swaths of land in prime areas..."
What a very shallow and pissant intent, unfortunately very Trump-like. I can easily imagine that, with his single dimensional thinking, he would try something like this.
Unfortunately, we've also got the tech-bros to contend with and the simmering manure-tea stew that is Project 2025.
Prof Krugman, you need to add "Catch me now I'm Falling" by the Kinks as a musical coda next.
Wyoming senator is trying (as did Utah but was turned down by the Supreme Court). Wants all public land except Yellowstone NP. Grand Teton is prime real estate but it is publicly owned along with much of the state. Prefer it stays that way.
Will voters ever get the hint?
No, because voters are still convinced Jimmy Carter started a recession.
I think that is super important and despite being erroneous is baked into consciousness that Dems are econ failures. Carter really wanted people in the US to invest in energy independence, stay the course for a correction to sustainability, to dismantle the CIA and proxy wars by the military industrial complex and instead we got 8 yrs of demented Reagan trickle down gold toilet seats on yachts and FAA Union busting all based on the the Iran contra coup.
Republicans are better at messaging than Democrats. And what’s funny is Republicans believe they are terrible at messaging because they are so invested in their asinine policies like tax cuts for the wealthy that only harm the regular wealthy while adding trillions to the debt. So if Republicans are so terrible at messaging…then why is Trump appointing so many Republican talking heads to his administration??
Not the ones who haven't already. There's a bit of buyer's remorse on the part of some who voted for Trump, but most are still true believers
Sadly, I believe you're right.
No.
Next question?
"There is a seeming disconnect between, on one side, the economic news (somewhat worrying) and the political news (terrifying) and, on the other, the stock market, which seems untroubled by all of it."
Consumer confidence has nosedived since Trump took office, as though people are surprised by all of this. Way too many voters believe Republicans in general and Trump in particular help the economy. It's the biggest lie of American politics.
Unfortunately Democrats like Jason Furman and Larry Summers are to blame by exaggerating the impact of the recent transitory inflation and minimizing the impact of inflation from 2005-2008. Republicans stick to talking points much better than Democrats as the people Trump has appointed talked up the Bush economy and then blamed Jimmy Carter and Clinton when the economy imploded in 2008. Democrats like Furman complained about the recent inflation and then after it has little negative impact make a big deal blaming Biden for it!?! Why don’t you focus on 2005-2008 inflationary period that peaked at 5.6% CPI in July 2008???
Obama was a mistake. We thought putting a black guy in was radical. Not only did we betray the New Deal, our fake radical has given us 20 years of white backlash and a WWC that is so blinded they can't even see they are destroying our democracy.
Very true except that the white backlash is more against wokeness and CRT. Obama made white people feel good. He made them feel broad-minded for accepting him, when he was really so unchallenging. Woke and CRT got in white people’s faces and told them they were racists and should feel guilty for their privilege. There was much for white people to examine and realize, but getting their defenses up was not the way to get through to them.
I am a centrist. I believe in (regulated) free markets. In that way I am just like Obama. The white-grievance right, I am quoting, "is anarchist." They believe in destroying our stability and our government. The Left, especially the black Left, is conservative. We want gradual progress from a stable base society. CRT is not radical. It is an academic THEORY that says racism permeates America's institutions and is self-reinforcing. Yes. Whites wrongly thought accepting Obama was radical and ended the race problem. It will take decades to fix our delusions that have given us Trump.
We need(ed) to be led by that black Left. (Too late now, at least for now.) The white Left fetishized race (thus continuing racism) and gender minorities. They needed those groups to be homogeneous groups, and to remain "marginalized," so they could be hero-allies. Voters preferred a less devious form of white supremacy.
Obama was also hamstrung by extreme republican stonewalling after losing control of the house after his first term. Specifically, them playing chicken with US government default via the debt ceiling to "starve the beast" (and damage the economy). One thing Dems did to try to get things moving is changing the senate rules so you can't filibuster appointees. They're without that tool now.
And then there's RBG...
Obama absolutely deserves blame. Quite a bit of it. But I'm not sure we could've really predicted how he would let us down when we elected him. Downside of electing someone with such a short record, I guess. I think Kamala would have been similarly hard to predict, had she won.
Politics is about WHY voters pick who they pick. We can talk about incompetent GOP dudes all day, but their voters are convinced and we need to stop ignoring what is really motivating them. They learned back in the Tea Party days to LIE if asked about race. We have to force the subject and we have to find the argument to convince them to quit....or the rich will play this dog whistle game until there is no democracy left.
not talking about the oppostion....well in this way: Obama made it easy for them to go to hicksville and holler "the black guy wants to give out welfare to undeserving minorities." And Obama was obsessed with showing them he was neutral on race. All that centrist crap he did was for nothing. Just like Clinton they hated him because blacks liked him and he was not a bend-over dog for the rich. Kamala followed his shut up about race game and Trump beat her because he was doing white grievance politics (aka dog whistling).
In the beginning, there was progressive growth, then it was followed by Conservative skimming, and the cycle has always continued.
Interesting comparison to the fracking boom. I would point out that there was also massive capital expenditures for the internet boom — just by different companies. Qwest Communications — and others — installed massive amounts of fiber optic cable, all of which was eventually used — but which led to their bankruptcy when the uptake wasn’t nearly as quick as they had anticipated. The fiber optic cable enabled much of the expansion in data usage through 2010 and beyond — but the companies that installed it had long since gone bankrupt.
Sure, but this assumes there's widespread adoption coming, like, ever. I have the same doubts that I do about Elon's obsession with self-driving cars: is there actually that big of a demand for it and what are the odds it ends up being a glorified "self-checkout" at the grocery store?
They are certainly trying to shove AI down everyone's throats. Every company that's heavily invested in AI touts their new AI products. I think they're desperately trying to justify their expenditures.
I see scientific uses like Alpha Fold, but do we really need to improve scammers and propagandists?
FWIW, I use AI a lot in my work (software development). While I hesitate to call it "transformative," it is certainly a technology that isn't going away, at least in my industry. Of course, one industry alone can't justify all the AI investments.
Yeah, that's where I am at. The Lexis AI I use at work (I'm a lawyer) is certainly useful, but it's so inaccurate that it's not really much more useful than a Lexis search. It can help you draft a brief, for instance, but you'll get laughed out of court if you actually submit an AI brief with no editing.
I find I resort to it to try and figure out solutions for problems after my normal methods at have failed, and I thus only use it on exactly the kinds of edge cases that it's unable to handle, I suppose because there's a lack of relevant high quality training data.
I should really flip this around and use it more for easy stuff, but my confidence in it is very low as a result of these experiences.
I was thinking along the same lines - there was quite a lot of capital expenditure during the internet boom. For another aspect, there was a lot of infrastructure bought and installed at the business user level. Servers, routers, desktop computers And laptops for everybody, etc. - which was not consumables, it had some lifespan.
Yes, the initial round of companies that failed were the ones that hade been valued on “eyeballs” or how many users they had; when they went bankrupt all of their equipment went up for sale in the secondary market and the NEXT year, the companies that had actually been profitable, with a sustainable business model, ran into trouble because the demand for their products (laptops, desktops, servers, routers, etc) declined sharply. Companies like Silicon Graphics disappeared, and companies like Cisco took decades to reach their prior valuations.
As a former emergency generator tech in NYC (all data operations have to have emergency power),I remember companies folding before their installations were even complete.
Btw, Lehman failed when BoE wouldn't countenance Barclays investment. Barclays later bought the wreckage, including a new data center with 6 2800kW generators (Lehman's 1st data center, 80s, only required 3 1000 kW). Oddly, the entire physical infrastructure cost Barckays no more than the contracts of key Lehman execs, iirc.
Do you think a similar phenomenon could happen with nuclear energy firms? The recent events drive an increase of SMV plants to match the expected increase of energy requirements for AI companies, but who end defaulting after making more than was necessary in the short run— the recent Microsoft closures of new data centers being particularly instructive here. The plants are the taken over only in the long run when alternative energy needs are required, but with the firms that built them long out— just like the fiber cables.
Gates has plans to build a small nuke (molten sodium) in western Wyoming now that the coal plant is on its last legs. Lot’s of speculation but maybe for not since EV charging station buildout (minus tesla’s) has been canceled. This project is/was suppose to save this small community.
Molten sodium. Dehydrated catastrophe. Just add water.
Was this the same plan he had before trump and had negotiated a deal with Xi to run these nuke pods in China but the trump admin was like "no" and reneged on the deal with Xi?
Have no idea.
The Professor and Mr. Chanos say the capital invested during the internet boom was bottom-up and not as intense and focused as the surge leading up to AI, which is being driven big companies taking the majority of risks in the hundreds of billions of dollars...nothing like Gates starting a company from his garage.
As a computer scientist I have always thought the AI boom is mostly a scam. If you talked to most people they probably think the Large Language Models which have gotten so much attention in the tech media are actually sentient (e.g. they have achieved general artificial intelligence). Nothing could be further from the truth. I see LLMs as really complicated probabilistic search engines which consume a huge amount of computing resources for their training and deployment. So far they have failed to meet the hype. Given the huge amount of investment being thrown at this technology, it is unclear whether the payoff will be worth the cost or even if there is a payoff.
As a behavioral neuroscientist I fully agree, LLMs are not even AI, let alone AGI, and it is infuriating that people have been duped by the tech bros who definitely know better, or should. The credulous news media have utterly failed to bring any skepticism to the completely unsubstantiated claims that AGI is just around the corner. There are plenty of experts out there who know that LLMs are not AI but they never quote or apparently even talk to them. They continue to call wrong answers "hallucinations" as if these programs have some sort of brain. The general public seems to believe that we are on the verge of SkyNet when what we really have is more like Teddy Ruxpin on ketamine.
I spent 20 years in data analytics and the issue no one ever solved was “dirty data”. All data sets have it and it makes doing data analysis difficult when you know that some of your data has errors. With AI, there is a huge dependency on feeding the system high quality data. But who decides what is high quality. If the system is fed bogus research that Ivermectin wards off COVID, the AI will think that is a valid solution.
If people who don’t know what they are talking about feed the system bad data, it will make bad decisions. In other words, there is real potential that some AI systems could just become a MAGA voter that can do faster math calculations.
Agree completely and would just add that it depends on who decides if data is "good" or "bad" and if the system can tell the difference or adapt to new information. Yes 1+1=2 but much of what is considered "good data" later becomes at least incomplete if not wholly incorrect, such as trans fats being better for you than saturated fats. Beyond that there is the entire realm of opinion which is not necessarily good or bad to say nothing about emotions and experience of the physical world through sensation and perception as well as social capacities like empathy. There are people who are incapable of empathy. Those people are often psychopaths.
We are far from AGI. Us computer types do tend to anthromorphize aspects of computing. I do find the "hallucinate" metaphor to be quite misleading. What is really troubling is we do not know why the LLM came up with a wrong answer. In a system with billions of compute noses it is pretty much impossible to trace down the caue of the problem.
Thank you for that clarification because I don't know this field but it smells like desperation and smoke and mirrors, yet there are threats of AI fueled mass layoffs coming shortly and AI insurance claim travesties that are fairly well documented, and the race to the bottom previously didn't depend on any technology working as well as it could/should have. I studied under Lester Milbrath re sustainability and AI back in the 80s and he predicted eventual collapse and decentralization - people literally going back to doing their own work, but right now idk if that's where we are headed.
As another computer scientist, I agree.
Thanks George. I spent 48 years in IT and have seen fads come and go. The probabalistic neural networks underlying the LLM tech is interesting and useful in some circumstances. But these systems are very expensive to train and run and the results are meiocre at best.
I personally think that there is a serious misconception about AI. There are many different approaches in AI and not all of them involve neural network. More importantly, there is not a all-encompassing AI model. In addition, to solve a problem, one does not have to use AI. A legacy application may actually be a better solution. Of course that may get one involved in a technology religion argument.
One AI model is one particular set or subset of knowledge (for lack of better terms ) in solving a problem. A knowledge domain can be considered as one particular, discipline, expertise or specialty. So Medicine is a domain. Within that domain, it can be further categorized into many subdomains: family medicine, cardiology, psychiatry, etc. The first step to build a successful model is knowledge acquisition. You have to know what problem you are going to solve, what objects and their relationship are involved and what the methods to solve the problem are. In neural network, this means the knowledge to build a training set. For instance, Language model is for language processing. The training set will require various language patterns, topics and contexts to understand the nuances of human language in order to generate the relevant responses for a question or a command. Furthermore, you cannot use a domain model for another domain. You cannot use an English model for Chinese interpretation. Even the vision model for a robot can be different from one for autonomous vehicle (vehicle in the other lane, its speed, etc). Government provides a lot of services, each of which requires different expertise (knowledge domains, e.g. taxation) and knowledge set (tax laws, accounting, etc). Do the Muskrats understand that (many of the things they did seems to indicate they do not) or does Musk simply lie to get his tax cut, to get the government contract for a multi-years and multi-billion dollars business, or all of the above. None of the new processes of a service can be developed in a short time frame and before they are developed, he disrupts or destroys many government services, infringes on people's privacy, destroy people's career, and potentially damages the countries' security.
There are many issues or challenges with AI, technically, economically, socially and in businesses in general. Without going further, I think the current events do highlight a problem. When a technology with such potential falls into the hand of some one like Musk who believes in a terrible and simplistic ideology, who has no moral compass, empathy and compassion for any one else, that can lead to disastrous consequences.
I agree there are different approaches to AI. Neural networks are but one approach. And as I said, there are domains where AI can be very useful as a tool. LLMs, however, are dominating the marketing for AI systems. Because LLMs exhibit seemingly human like behavior when it comes to written language, many folks think they are sentient and have reasoning ability. They are in fact only as good as their training sets. They are reasonably good stochastic parrots.
I agree that an application is only as good as its training sets. Important thing is that training sets can be modified and improved.
The more sophisticated applications involve the integration of multiple domains. One particular use case is in healthcare for diagnosis. Using computer vision, it can analyze a medical image such as a X-ray and feeds the output to a LLM to correlate these findings with patient history and medical research, delivering diagnostics and potential treatment options. Other noteworthy use cases are surveillance, inventory control and quality control. I do not know at what stage these application are, but they sound promising.
One thing I also learn from past experience is that we can never under-estimate the salesman of, say, Microsoft. I have seen them successfully sold very terrible products.
I treat the current rush into AI as the infrastructure building phase. Of course that will bring its own problems and issues. Along the way, there will be more confusion and chaos but there will also be technology advances. We will see changes in computer networking due to wide bandwidth requirements, environmental impact, such as using nuclear power, and advances in computer platform, such as quantum computers. At this point, I am more concerned about its abusive uses and the human cost these will bring.
You are only seeing what is at the end of you nose.
AI for intellectual tasks is limited. But AI + robotics promises eventually to replace nearly all human labor
That will lead to an immense revolution in how we live. The sky is not the limit, it's just the starting point.
On what evidence do you make this assertion?
On what evidence did Einstein imagine spacetime? I think he just used his noodle. And that was some noodle!
Rema, Not to find fault with your news gathering, but isn’t this supposed to be about comments on the Krugman posts?
It kinda feels to me like you are posting to gain subs to your stack.
I guess you neither read my comment, the news, nor my name just above your comment?
I guess not 🤷🏿♂️
Rima Regas,
You are consistently a good read. Your experience and brains have given you remarkable and extremely valuable insights. I hope you will continue to share your take on the passing scene with the public.
Thank you so much, James! I will do my very best to keep up the commentary and the news gathering. There is so much going on!
Honestly I think the boomers stayed healthy longer than most predicted and kept Joanne fabrics alive….gen X is probably sewing things but they will buy everything from Amazon. I can’t believe Barnes and Noble still exists when I go to the grocery store!! WTF???
Have you gone inside to see if there are shoppers there browsing books and having coffee?
No, and it used to be my favorite place to go.
Same here.
As a semi-related side note to Paul's analogy, it takes a Yooooge amount of water to frack an oil well.
There is no "typical" well, thus the gallons of water/well ranges from 1,500,000 in the Bakken to 15,800,000 in the Horn River Shale in BC, CA. Most fracking water now comes from surface sources that, increasingly, are augmented by deep groundwater as drought impacts much of our nation, particularly the big fracking states. Big Oil is relying on more and deeper wells to access ancient aquifer water to force decreasing amounts of oil out of each fracked o/g well.
While the total amount of water used for fracking is relatively small compared to that used by, say, notoriously inefficient Big Ag, (BTW, much of Big Ag's agua is used to grow corn/beans to produce low-quality ethanol to burn in our 4wd, urban SUV/Pickup Fuelhogs), we still commit an estimated 125,000,000,000 gallons of our decreasing water supplies to frack wells for fossil fuels that we burn and then dump the resultant GHGs into the atmosphere of Our Only Home. Globally, we created a record amount of GHGs in 2024 (Yay, us?) - but that's a different topic.
Estimates vary widely, but, roughly, we commit enough water to supply 5,000,000 US homes/yr to frack wells for more oil/gas to burn - and further damage our environment. And, you might ask, what happens to the water that was used for fracking? Well, dang, turns out that those billions of gallons of fracking water are toxic. Disposing of that toxic water is a another Yooooge problem that results from our Fossil Fuel Addiction. Much info online about that disposal prob.
Here's one related article: www.nytimes.com/interactive/2023/09/25/climate/fracking-oil-gas-wells-water.html
More Perfect Union had a great video on the dangers of toxic fracking waste.
https://youtu.be/J4tXtSFM5NE?si=XHc9QwHOyv2N_3sa
I think the word you are looking for is denial. Market denial. 🤯
We set ourselves up for this, though. As a society we have determined that any activity that generates financial gain is inherently good. The markets are not in denial, they are taking full advantage of the chaotic situation. Eventually they will take their profits to Aruba and leave the rest of us to enjoy the smoking ruin our economy will become. And Republicans will find a way to blame it on Democrats
This is what Mark Andreesen more or less stated in his poorly written manifesto. The billionaire class thinks growth at any cost is good. And, as I often like to point out, while Andreesen is a billionaire, he is not very smart.
Isn't that the same thing as "willful ignorance"?
I am not particularly knowledgeable about economics, but I do have pretty long experience in boom and bust cycles in support for research in machine intelligence (a more inclusive term historically than AI). And that experience, which has its correlate in the commercial world, tells me that we should expect that technological successes will lead to unreasonable expectations, and that those expectations will not be realized and the wave of support will subside. Along the way, we will also be surprised by some applications working better than we expected (such as the large language models that have taken so many by surprise starting a few years ago). But overall many will think the technology can do much more than it really can.
Also, because of the huge costs involved in training (and drawing inferences from) the huge models, there will be major efforts to find cheaper/smaller ways to get similar results, which will limit the expansion of hardware requirements.
The apparent market calm (before the storm) sounds like a toned down version of "irrational exuberance".
The increase in inflation expectations is no surprise. Sticky prices plus inflationary policies are a volatile combination. Interesting that all the layoffs - both public and private - don't appear to have influenced this - yet.
Its only been a few weeks.
My son who fights wildfires and does fuel reduction in the off season has no work. The Federal grant money isn't coming through.
Wow, that sucks - for him and for all of us. I'm not looking forward to the next fire season, it's going to be devastating.
Prepare for the great collapse. Musk and agency heads are at odds. Trump has no plans just bullying. Macron stood up to him, we are disconnecting from the world. Does anyone think that is a good thing?
Was thinking about moving to France... which is looking better after seeing Macron step up. Bon!
You assume a US citizen with U$D will just easily be able to transfer that money there for Euros and be automatically granted residency. Why would France want another free health care recipient and accompanying U$Ds?
I know people who think it’s a great thing.
Freud coined the word “compartmentalization” to describe how many deal with cognitive dissonance. Sounds like the players Chanos describes and when you overlay “gaslighting” ie emotional fracking, looks like 2025. voila as they say.
The nub with the fracking companies is that they funded with high yield fixed rate debt to prevent equity dilution. This meant that they couldn’t leave the oil in the ground when the prices tanked. When the price drops the more they pumped the less profitable they became. New financings came with restrictions that allowed a moratorium on production based on market conditions. This is why the “drill baby drill” crowd needs to be circumspect.
An additional challenge came from oil rich countries that had lower production costs where they could just turn on the spigot and flood the market with excess oil.
Isn't there in fracking less technical flexibility to "crank down" for fear of losing "recoverability"?
Correct, therein lies the problem. It is like they are addicted to frack.
But isn't there an effective monopoly consolidation here in the US among producers and price fixing non-compete with OPEC? Read something last yr re this mvmt around pricing/competition. Idk how the financing part fits in with this
I was referring to the first two crashes that I was aware of when I was in commercial banking and corporate finance.
After the crash of 2020, I do recall that there was talk among Trump, OPEC, the Russians and the frackers to limit production to keep the prices high. Hence, the runup in prices under Biden.
So, I think that there was some sort of collusion.
That we even have to clarify for each other which of several crashes in <20 yrs is what's wrong with this picture. incompetent asshats and unbridled privatization, tax cuts and profiteering. I'm not even a masters level and could see both the first and second coming, the whole feigning surprise by gvmt leadership in 2009 was infuriating.
Trump is the first president to win because oil prices were too low in 2016 and it caused a mini recession in Rust Belt states because they make fracking equipment…and then he won in 2024 because oil prices were too high not in 2024 but in 2022!?! And so 60 Minutes had a PA diner owner that voted for Trump complaining about oil prices when high oil prices are actually good for her business!?! America is the biggest oil and gas producer by far!?!
But that’s how you know Republicans are better at messaging than Democrats because they can manipulate every issue to their advantage…although having voters treat politics like football fandom helps a lot! Although most football fans know more about their team’s history than Republicans know about the current issues.
omg, these musical codas are making me feel old! i remember sitting in the theater watching this very young David Byrne give it all - can't believe that was more than 40 years ago!
I started law school in 1981, and still remember seeing a meme in 2021 that said, "Friends don't tell friends that 1984 was 40 years ago." When I saw the meme, my reaction was, "WTF, that can't be right!" (Sadly, it was, which meant that my "study group" and I went to our 40th reunion last year.)
How funny! I was in law school in 1984 too. I started in 1982, so my 40 year reunion is this year. My daughters have been referring to the 80's as the dark ages for years, making me feel like a dinosaur when I was only in my 40's...!
Excellent and convincing connection to fracking. Pull back by Mr. Softie first warning sign. Thanks Paul for your ongoing efforts to infuse sanity into the public discourse in the US
No, inapplicable and misleading. Fracking is an extractive activity. And all such activities peter out. Oil fields deplete, gold mines become uneconomical, whaling destroyed itself, etc. But car manufacturing only got bigger with time, and no reversion to the mean.
(sigh), the comparison was not AI as a commodity to oil...the comparison was on expected return. In fracking's case people expected too high a return and there was too much money chasing too small a profit. That is how Mr. Chaos expects AI to play out.
It’s an analogy, and none are perfect, otherwise we wouldn’t need them.
Your car analogy is even more flawed. They are way more durable than AI-enabling computer chips.
The older ones are. The new ones are packed with computer chips. When my wife bought a new car last year, we were told that it would cost $2,000 to repair/replace the computer if it started giving trouble.
That’s true, but it doesn’t change the fact that a car depreciates much more slowly than an AI enabling computer chip.
In a sense AI is extractive, being dependent on all the (digital) knowledge that exists. As I see it, there just isn't a lot of new knowledge being found.
There are scientific studies where the huge computing power really does make a difference. Large data sets allow patterns to be seen.
I think the problem is AI is being sold as the be all end all for everyone instead of the scientific institutions that it will really benefit.
I don't want AI to set up a vacation for me or plan my wedding or decide where I should go to dinner tonight. I don't want it to write my stories for me or draw pictures for me.
I would like it to do the dishes and sweep the floor.
Yeah, right. Time for Musk to close the Patent Office.
Oh, wait. Did you see about successfully using CRISPR to remove Downs Syndrome extra chromosomes from stem cells? 25th Century science.
AI is comparable to fracking or streaming services in that capital assets (chips, wells, proprietary content) depreciate fast - faster than the 20%pa used for accounting purposes, meaning that the reported profits may be completely overstated to non existent.
Industrial robots used in auto manufacturing have an annual depreciation of around 15-20%, fracking wells loose about 40-60% output in one year, AI chips depreciate due to the sheer scale of technological progress (50% pa?)
Undermining confidence in the economy is easy to do and slinks along through the grass until it springs up to bite.
Things to consider:
Half of the US is terrified that Trump/Musk are driving us off a cliff domestically and internationally. How many of us (the most educated and affluent) are going to start up or expand businesses with that going on? Half of America is slamming on the brakes and shutting their wallets to hunker down for bad times to come.
The Federal government has all kinds of tributaries flowing through the economy. When it suddenly dries up people are left high and dry. In my area the people who do forest project work for fuel reduction aren't working. We wonder if there will be firefighters when the summer wildfires begin. The county government wants to lay off a big percentage of their workers because they say money won't be filtering down from the Feds. The effects are just starting to ripple through the economy.
Jim Chanos' perspective is very illuminating, thanks for introducing us to him. The fracking bubble does fit better in many ways.
Also, wanted to share 2 quick thoughts:
1) You, Jim Chanos and someone like Mark Cuban really ought to do a joint podcast to talk about the state of the economy and markets. The 3 of you would be interesting to listen to. Not a regularly scheduled thing, just something to do whenever the 3 of you could triangulate it.
2) Some weekend I hope you can share with us about how Isaac Asimov's "Foundation" novels influenced your career choices. Sci-fi and economics may seem like unlikely companions, unless you've read Asimov. 😁
Thanks!
Ed Zitron has been yelling about this for a few years now https://www.wheresyoured.at/longcon/
He ties it into a larger picture of the 'Rot Economy' in Silicon Valley; that the singular drive to monetize EVERYTHING in IT has steadily degraded the very utility the internet wrought. Things like Google deliberately degrading their search by focusing on increasing their revenue by pushing paid content over what you actually searched for.
If you feel that your 'google-fu' has gotten worse, that you're spending more time trying to find an actual answer to your question, that you're wading through things that aren't useful to your question...IT'S NOT YOU.
And AI isn't helping this materially because AI very very much has a GIGO problem.
And all of this is before reckoning with the fact that AI is hugely expensive, without any clear way to turn that expense into something actually useful.
I quit Google for most search. I use Duck Duck Go instead. Saves me a lot of time because Google is basically a giant advertisement platform.