If you’re feeling an emptiness inside you where the first-Friday-of-the-month jobs data would normally be, know that you’re not alone. We’ll get through this together.
And breaking late Friday, Bloomberg reports “China is pushing the Trump administration to roll back national-security restrictions on Chinese deals in the US, dangling the prospect of a massive investment package as part of a proposal that would upend a decade of policy. ... The Chinese floated a figure of $1 trillion earlier this year, according to one of the people, but the size of the potential investment being discussed now is unclear.” There are scant details, and it’s probably too early to determine how serious a proposal this might be in the negotiations with China, but suffice to say, this would be a very, very bad idea.
This week is Oren’s turn on the soapbox and he has a few things he’d like to say about AI slop:
It was just last week that OpenAI CEO Sam Altman, apparently out of cash and now giving Nvidia $100 billion of equity in exchange for more of those sweet, sweet GPUs, explained on CNBC:
We’re so limited right now in the services we can offer, there’s so much more demand than what we can do. And as we look forward another year or two years, let’s say it takes 10 gigawatts of compute, 5 gigawatts of compute, you could choose one of two things: you could choose to cure cancer by doing a bunch of heavy AI- research or you could choose to offer free education to everybody on earth. No one wants to make that choice.
So imagine my excitement this week when OpenAI announced its next major product release. Which was it doing first—curing cancer, or educating the world for free?
Oh.
Sora 2 lets you make videos of a figure skater dancing with a cat or a man doing a backflip on a paddleboard (yes, those are OpenAI’s first two examples) and share them in an addictive infinite-scroll app designed to compete with TikTok. “It’s the most powerful imagination engine ever built,” explains fake-AI-generated-Sam-Altman in the creepy intro. Well, no, Sam, whatever generates your talking points about curing cancer or educating the world for free next year surely holds claim to that title.
I guess, in a sense, we really are looking at the most powerful imagination engine ever built… (Photo credit: screen capture from OpenAI Sora 2 introduction.)
Sora 2 is designed around the idea of promoting deepfakes as fun, cool, and natural. “By observing a video of one of our teammates, the model can insert them into any Sora-generated environment with an accurate portrayal of appearance and voice. This capability is very general, and works for any human, animal or object.”
Maybe you thought that would be reason for concern, but, “we first started playing with this ‘upload yourself’ feature several months ago on the Sora team, and we all had a blast with it. It kind of felt like a natural evolution of communication—from text messages to emojis to voice notes to this.” What could be more natural than [checks notes] making a video of yourself playing the trumpet in the middle of a stampede of zebras?
It’s not antisocial, either. In fact, “last week, we launched the app internally to all of OpenAI. We’ve already heard from our colleagues that they’re making new friends at the company because of the feature.” I am trying to think of another example where a company took serious concerns about the unintended consequences of its product—in this case, the proliferation of deepfakes, loss of connection to reality, ease of manipulation and addiction—and just released them as a wonderful new set of features. I’m coming up empty. We never had “death cigarettes, which help you make friends on the lung cancer ward.”
At least, to OpenAI’s credit, they’re not looking to make money on this, like those other, gross social media companies (Meta released its own version of the same product last week). You see, “a lot of problems with other apps stem from the monetization model incentivizing decisions that are at odds with user wellbeing.” Not OpenAI. “Transparently, our only current plan is to eventually give users the option” [an option? to do what?] “to pay some amount” [hmm, so monetization?] “to generate an extra video” [so, the exact thing you’re trying to get users hooked on doing?] “if there’s too much demand relative to available compute.” Ah, whew.
Is too much demand relative to available compute a foreseeable challenge for OpenAI? If we were choosing between curing cancer, educating the world, and making cat videos it might be. I guess we’re lucky the models have only shown even the slightest ability to do one of the three.
As if Sora weren’t bad enough, friend-of-the-Stack Tyler Cowen swooped in yesterday at The Free Press with one of the more bizarre columns you’re likely to find in a mainstream publication this year, “My Favorite Actress Is Not Human: Tilly Norwood doesn’t need a hairstylist, has no regrettable posts, and if you wish to see a virgin on-screen, this is one of your better chances. That’s because she’s AI.”
The column is superficially embarrassing in all sorts of ways that have already generated much-deserved ridicule, but it’s worth plumbing a bit deeper to understand the connection between the kind of thinking on display here and the kind of thinking that gives us Sora. Cowen is, after all, “Silicon Valley’s favorite economist.” If you find yourself wondering, “what could they possibly be thinking?!”, this is a good place to look for an answer.
For instance, Cowen has an abiding faith in perfectly efficient, competitive markets and perfectly rational, welfare-maximizing individuals to deliver optimal outcomes. Thus, he condescendingly mocks the Screen Actors Guild because “in the same breath they insist that a) Tilly will put human actors out of work and also that b) audiences do not want the product.” A contradiction! How could some new, cheaper mode of production ever gain traction if it does not align with true consumer preference? “If that is the state of your logical thinking these days, you have problems well and beyond competition from AI.”
Except, of course, that we have seen repeatedly in recent decades the way new waves of technology have made people actively less happy, especially in the digital realm, and degraded the quality of products and experiences, especially when labor costs can be saved. One person concerned about this is Tyler Cowen, who writes just two paragraphs later, “currently I think Hollywood is failing us; the quality of its movies has never been lower in my lifetime. Most of the top hits are boring and predictable tentpole franchises. Fight and chase scenes are overdone and laden with CGI at the expense of good dialogue and dramatic content.” But only a moron would think that the opportunity to more aggressively substitute the human elements of filmmaking with technology could be cause for concern.
I suppose the quality of the case should rest on whether Tilly is in fact a good actress. Is she able to convey human emotion especially well, or deliver especially sharp comedic timing? I don’t know, and neither does Cowen, because there are no meaningful examples of her acting—just Instagram posts. Indeed, the technology to make a movie with her doesn’t yet exist. That’s why, tellingly, Cowen’s list of attributes he likes about her have nothing to do with acting, and instead come down to the objectification of actresses as high-maintenance, looks-obsessed public relations liabilities with undesirable sexual histories.
Not Tilly. “She is beautiful, but not too intimidating. She has a natural smile, and is just the right amount of British—a touch exotic but still familiar with her posh accent. Her Instagram has immaculate standards of presentation.” (What does it say about someone if these are the criteria for his favorite actress?) But wait, there’s more:
I suspect that Tilly is easy to manage, has no past political posts to regret, and if you wish to see a virgin on-screen, this is one of your better chances. She needs neither a hairstylist nor a sushi lunch, and she is not going to walk off the set because someone offended her. Tilly will stay young forever—at least if her followers want her to.
Let’s be clear, none of this has anything to do with making great art, which Cowen professes is his real priority. Frankly, it’s perverted, which is one of those terms we’re not supposed to throw around in polite company, but I’m not sure what else to say about someone who sees an AI-generated posh young British avatar and thinks, I really like that she’s never had sex and will stay young forever. For the record, an AI avatar is not a virgin any more than Buzz Lightyear is a virgin. It’s a nonsensical construct. Also for the record, if you’re watching an actress on screen, and thinking, “hmm, this is a good performance, but I’d like it better if she had never had sex in real life, that’s really getting in the way here,” you have problems well and beyond competition from AI.
The case that Tyler ultimately makes here, the case that OpenAI ultimately makes for Sora, is that all this comes down simply to individual choice. “Many observers hate short-form video, which they consider to be a kind of slop that ruins our minds,” acknowledges Cowen. “I do not dismiss such claims, but I will proudly report I will not let it wreck me, just as I currently avoid TikTok.” (You’ll recall, Cowen also wonders, “if screen time is making kids so miserable, why won’t they seek out methods to make their screen time more efficient?”) OpenAI likewise assures, “protecting the wellbeing of teens is important to us. We are putting in default limits on how many generations teens can see per day in the feed,” which teens will surely leave in place, because teens love limits generally.
As for slopification of the culture generally, Cowen advises, “learn how to rise above it. In other words, take responsibility for your own level of taste. If you feel like you can talk about slop, you also ought to be able to avoid it.” This is not just someone who seems incapable of comprehending how culture works, this is someone who actively dislikes and devalues its very existence. Cowen avoids slop, and is going to use these incredible new tools to make his own short-form videos that will surely be great works of art viewed by several dozen other people. As for the shared culture in which the little people live and upon which they depend? Too bad.
This is the economics of Silicon Valley. It is selfish and it is dangerous. It ignores reality, history, and human nature, preferring instead a blind faith that whatever can be invented will surely redound to the benefit of all, just give it time. The people through their democratic processes cannot possibly intrude upon the prerogatives of their betters who write the code. “If Tilly is the one to shake up Hollywood and induce higher quality standards—or something on the screen that is indeed truly human and humane—that could prove her greatest achievement,” Cowen muses. If, indeed. Does that come before or after Sam Altman cures cancer? With a few more gigawatts, perhaps we can do them both next year. —Oren
YET MORE ON THE AI REVOLUTION THAT ISN’T
In other AI news, the Wall Street Journal went big with a classic dog-bites-man story, “Walmart CEO Issues Wake-Up Call: ‘AI Is Going to Change Literally Every Job.’ Head count expected to stay flat over next three years, despite growth plans, as AI eliminates or transforms roles.” Not mentioned: headcount also remained flat, or even declined, over the past three years, and the three years before that… in fact, it is flat back to 2010.
The Harvard Business Review reports, “AI-Generated ‘Workslop’ Is Destroying Productivity.”
Employees are using AI tools to create low-effort, passable looking work that ends up creating more work for their coworkers. … We define workslop as AI generated work content that masquerades as good work, but lacks the substance to meaningfully advance a given task.
And keep an eye on “griefbots” and other efforts to resurrect the dead by creating chatbots trained on behavior during their lives. From the Financial Times, “AI Can’t Raise the Dead, But It Might Do the Next Best Thing.” CNN has featured the avatar of a school-shooting victim created by his parents. Who thought it would be a good idea to put that on TV. Consumer welfare maximization in action?
For a better perspective, read Pope Leo XIV on his chosen theme for World Communications Day, Preserving Human Voices and Faces: “While these tools offer efficiency and reach, they cannot replace the uniquely human capacities for empathy, ethics and moral responsibility.”
GOOD WEEKEND READS
In Vanity Fair, James Pogue profiles Zohran Mamdani. NYC’s mayoral candidate has Kennedy-like charisma, a global profile, and nepo baby instincts. He is also a proud democratic socialist who has both Donald Trump and the left-wing establishment in a lather. Is Zohran Mamdani the future of American politics—or a fantasy?
In Engelsberg Ideas, Graeme Thompson recounts The End of Pax Britannica. A mercantilist, multipolar geopolitical system appears to be emerging from within the liberal world order, destined to replace it. The experience of Britain’s imperial past suggests that this transformation is unlikely to be peaceful.
IMMIGRATION ENFORCEMENT’S INTENDED CONSEQUENCES
The New York Times, reporting on how “Trump’s $100,000 Visa Upends Lives,” acknowledges “The changes are expected to force companies, which typically pay much of the cost of an H-1B visa, to hire fewer foreigners and to be choosier about whom they hire. That could, in turn, create demand for American workers and push up wages.”
The Wall Street Journal, reporting on how “Miami Suburb’s Once-Vibrant Housing Scene Is Hit by Exodus of Migrants,” discovers that “rents in Doral have dropped to their lowest level in three years.”
Conversely, Federal Reserve governor Stephen Miran highlights new research that finds large effects of increased migration on rental prices in Denmark.
THE ROBOT RACE
In case you missed it last week, the new American Compass collection, For Whom the Machine Toils, is very much worth a perusal, with a range of essays, discussion, and polling data on the relationship between technology and labor. Are American workers’ interests and American techno-industrial strength in tension? Many say yes, from labor activists who treat technological innovation as a threat, to libertarians who view worker power as an obstacle to technological dynamism, to economists who say that upholding labor standards and protecting domestic industry are both misguided.
The American Compass project argues they are wrong. An economy in which worker power and industrial power reinforce each other is possible. Rapid technological progress is precisely the formula for increasing both productivity and wages. Worker power, properly deployed, forces capital to invest accordingly, improves the return on that investment, and ensures labor enjoys its fair share. Protection of the domestic market—both workers and industry—was a central tenet of American economic strategy for decades.
But this arrangement is not guaranteed. In the face of rapid and disruptive technological change, including the advent of artificial intelligence, policymakers must grapple with these tensions—especially amid public anxiety about technological advancement’s effect on workers, even as American policy seeks to reclaim technological leadership and revitalize American industry.
Read Oren’s essay, “Two Cheers for Automation,” on why we desperately need productivity growth, and an economy that gives workers the gains. And then read Michael Lind’s essay, “High Wages and Technological Innovation: There Is No Alternative,” on why, if you want your country to flourish in the modern world, it needs to adopt a high-wage, high-tech model.
Keep those ideas in mind as you read stories like this one from the New York Times, “There Are More Robots Working in China Than the Rest of the World Combined.” Factories in China installed ten times more robots than factories in the United States last year, which is a disaster. As Andreessen Horowitz explains, “America Cannot Lose the Robotics Race.”
SPEAKING OF CHINA
Jensen Huang may think that being a China Hawk is “a badge of shame” and that the CCP should give Chinese firms “as much support as they like–it’s all their prerogative,” but this new NBER paper, “The Dynamics of Technology Transfer: Multinational Investment in China and Rising Global Competition,” confirms what everyone else already knew:
US multinationals formed joint ventures in China for market access and lower labor costs. However, these ventures transfer technology to Chinese firms, fueling future competition. While individual firms weigh the risks to their own profits, they disregard the negative impact on other US firms and the broader economy, resulting in an over-investment that may reduce the US welfare. In our empirical analysis, industries with more joint ventures in China show positive spillovers to Chinese firms but negative outcomes for firms in the US.
Proud Badge-of-Shame wearer Robert Lighthizer, U.S. Trade Representative in Trump’s first term, was in Mexico last week delivering a tough message (per Bloomberg):
Mexico has already made a first concession, announcing tariffs unilaterally earlier this month on imports from countries without a free trade agreement — effectively targeting Chinese shipments. But in the current climate of superpower animosity, that is unlikely to satisfy a Washington requiring complete alignment. President Claudia Sheinbaum will have to decide how far she is willing to go in confronting the world’s second-largest economy — and key supplier of inputs and components for Mexico’s factories — to keep Trump happy. … [A]rguments that China uses Mexico to reroute trade into the US gain at least some mathematical support. The problem is that, while Mexican business leaders would be delighted to buy more American equipment and machinery to narrow the gap, it was the US that offshored many of these supply chains to China and Southeast Asia in the first place.
Concerned by Mexico’s tightening alignment with the U.S., meanwhile, “China Launches Sweeping Probe into Mexican Tariffs as Trade War Widens” (FT). But with EU leaders still wavering in the middle, “China sends 2,000 workers to build battery power in Europe” (also FT):
China is locking in European dependence on its technology by sending thousands of workers to build cutting-edge car battery factories that the continent needs to breathe new life into its auto industry.…“I don’t think the Chinese want to share the knowhow with us,” said José Juan Arceiz, a Stellantis employee who is secretary of the company’s European works council.
For more scorching bad takes on China, check out Carl Benedikt Frey in Foreign Affairs, suggesting we should just outcompete China the way we outcompeted Japan, and Martin Sandbu in the Financial Times, suggesting that global trade imbalances might be just fine because the United States ran a large export surplus as it rebuilt Europe after World War II, and that worked out well for Europe. We may need a total and complete shutdown of bad historical analogies until everyone figures out what’s actually going on.
AND SPEAKING OF BAD TAKES…
In case you missed it, the United Kingdom’s National Health Service notes the “benefits” of first-cousin marriage, like “stronger extended family support systems and economic advantages.”
The Chicago Teachers Union celebrates the life of convicted cop-killer and Most Wanted Terrorist Assata Shakur.
And over at The Argument, introducing its first podcast, editor-in-chief Jerusalem Demsas lists “everything we care about—economic growth, abortion, immigration, free speech,” which summarizes perfectly the failure of center-left neoliberalism. “Grow the economy without having kids by importing foreigners and then talking about it… ideally over brunch.”
FINALLY, SOME INTERESTING SHIFTS IN THE POLLING DATA
Two new polls from Pew caught our eye:
Growing share of Americans say fewer people having kids would negatively impact the U.S.: “Over half of Americans (53%) now say fewer people choosing to have children in the future would negatively impact the United States. This share is up 6 percentage points since last year.
Americans increasingly see legal sports betting as a bad thing for society and sports: “Today, 43% of U.S. adults say the fact that sports betting is now legal in much of the country is a bad thing for society. That’s up from 34% in 2022.” Of particular note, “47% of men under 30 say legal sports betting is a bad thing for society, up from 22% who said this in 2022.”
Enjoy the weekend!
Here is an AI data point. I am researching buying a generator for camping. I camp anywhere from sea level to 10,000'. Generators suffer from a condition known as derating whereby they lose about 3% of their power for every 1000' above 1000' plus incurring some safety and fuel efficiency effects. This is caused by the lack of air at higher elevation and consequent problems with the fuel air mix. You can partially get around some of this by putting different jets in the carburetor. All this was easy to figure out but when I tried to figure what happens when you come back to sea level with the high altitude jets, I hit a wall, so I asked AI. The answer I got seemed reverse of the logic I expected about the fuel air mix. When I challenged the answer, it reversed itself and said I was correct. So did it do further research to reach the revised answer or was it telling me what it thought I wanted to hear. Who knows? (If anyone cares about the generator question, the answer is possible generator damage running a high altitude modified unit at sea level.)
First, another well written piece. Second, remind me to never piss Oren off. My goodness, Tyler Cowen may well be convalescing in a far-away hidey-hole after that beat-down. I’m not a fan of his, find some of his writing intellectually weak and even annoying, but I felt badly for him after reading Oren’s dismantlement of the poor intellectual ninny. Ouch!