🤖 The Age of AI vs. the Age of Populism
Is Washington ready for the challenge?
“Our bodies are made of stardust, and the atoms that make up our bodies were forged in the hearts of stars. From the tiniest subatomic particle to the vast expanse of the universe, there's beauty and wonder everywhere. We are all connected.” - NASA Administrator Bill Nelson
🤖 The Age of AI vs. the Age of Populism
I don’t write much about the Federal Reserve or monetary policy in this newsletter. But there was a tense moment Tuesday during Fed Chair Jerome Powell’s regular testimony before the Senate Banking Committee, a moment I think is relevant to the Faster, Please! ethos. After Powell said continuing high inflation and labor shortages could force the central bank to raise interest rates higher and faster than they expected, Senator Elizabeth Warren went on the attack, telling Powell that he was “gambling with people's lives” and painting him a heartless, green-eyeshade banker:
Warren: Chair Powell, if you could speak directly to the 2 million hardworking people who have decent jobs today who you’re planning to get fired over the next year, what would you say to them? How would you explain your view that they need to lose their jobs?
Powell: I would explain to people more broadly that inflation is extremely high, and it’s hurting the working people of this country badly — all of them. Not just 2 million, but all of them are suffering under high inflation, and we are taking the only measures we have to bring inflation down.
Warren: And putting 2 million people out of work is just part of the cost, and they just have to bear it?
Powell: Will working people be better off, if we just walk away from our jobs and inflation remains 5, 6 percent?
Whether or not you think Powell is doing a good job, Warren is wrong to unnecessarily turn a complicated economic policy debate into a morality play. I mean, it’s hardly unreasonable for Powell to look at the economic data and conclude higher rates are needed or that sustained higher inflation would be incredibly damaging to the US economy.
Interregnum: High sustained inflation a) erodes the purchasing power of wages and reduces living standards, b) makes it harder for families to plan and save for the future, c) creates uncertainty and volatility in financial markets and discourages investment and innovation, d) increases borrowing costs and debt burdens for businesses and households, e) undermines the credibility and effectiveness of monetary policy and fiscal policy. None of this should be particularly controversial. Total Econ 101 stuff, and I use that description in the most positive light possible.
Warren, however, was playing the role of fiery populist politician, and populists aren’t big on a) nuance, b) assuming good-faith efforts by opponents, or c) taking into account constraints and trade-offs. And while Warren’s performance should have zero effect on America’s independent central bank, we’re going to need better from our politicians in the Age of AI when along with plenty of creation there’s going to necessarily be destruction. But will we fear the latter so much — egged on by populist opinion makers — that we don’t accept the former and thus miss out on its benefits? I can imagine a different sort of a congressional hearing in the not-so-distant future:
Populist Senator: Could you speak directly to the millions of hard-working American writers, artists, teachers, and lawyers — among others — who you’re planning on getting fired through the release of TalkGPT 5.0? How would you explain your view that they need to lose their job in pursuit of faster productivity growth?
Tech CEO: I would explain to people that with progress always comes disruption. But history suggests that America tomorrow will be better than today thanks to what economists call “creative destruction.” It’s easy and natural to empathize with someone losing a job today due to technological change. But the alternative is a false security since without change America will be less innovative and competitive tomorrow, both economically and militarily. Lots of the jobs our children will work at haven’t been invented yet. TalkGPT will cost some jobs, but it will also make many of us more productive at what we do, as well as create new jobs. That’s the important lesson of economic history.
Populist Senator: And putting millions of people out of work is just part of the cost, and they just have to bear it? We need to tax AI and robots and we need to tax them now! Also, I’m running for president!
I don’t think this is a far-fetched scenario given there’s already been talk about robot taxes (Bill Gates, Bernie Sanders), and the populist right seems to have little tolerance for the disruption that is inherent in dynamic, entrepreneurial market capitalism. Venture capitalist Marc Andreessen anticipates such pushback, which is why he’s not worried about massive AI-driven technological unemployment. Andreessen argues that AI and other innovation is already “illegal” for most of the American economy. Sectors like education, health care, and housing are dominated by government and industry barriers that prevent technological innovation and drive up prices. The standard arguments against AI-driven unemployment are not even needed, he concludes:
Also: None of us can know where this is leading, how much disruption and how quickly. I’ve previously noted that the Metaculus prediction community has significantly moved up its consensus dates for human-level artificial general intelligence, to 2040 now versus 2059 a year ago.
I mean, it certainly gets my attention when Wall Street research starts mentioning AGI. This from Citigroup:
The ultimate form of AI, artificial general intelligence, should, in principle, be able to learn general problems. In recent years, LLMs have increased exponentially in size, with some describing them as increasing by ten-fold every year. This has led some to question whether the development of LLMs could be considered as a new Moore’s Law. Saying that, with the increase in the size of LLMs, there has been an increase in computing power needed to train them, so much so that the capabilities of AI are beginning to be largely restricted by the computing power currently available. With a single NVIDIA V100 (a GPU specially designed for AI training), it would take 355 years and a $4.6 million electricity bill to train GPT-3 to produce human-like text. The next iteration, GPT-4, will allegedly have 100 trillion parameters, 500 times more then GPT-3. With Moore’s Law under pressure in recent years, training GPT-4 models will likely be an extremely difficult task for even our best computers.
For now, the American economy needs more advanced tech in its economy — more AI, more robots — to replace humans doing some things, to help humans do other things better and giving us more time to do them, and to create new things for us to do. We’re not yet at the Technological Singularity. Even when marvelous new tech is out there, it can take a long time for business to adopt it. From The Economist:
“I think we might exceed a one-to-one ratio of humanoid robots to humans,” Elon Musk declared on March 1st. Coming from the self-styled technoking of Tesla, it was not so much a prediction as a promise. Mr Musk’s car company is developing one such artificially intelligent automaton, codenamed Optimus, for use at home and in the factory. …Given that Mr Musk did not elaborate how—or when—you get from a promotional clip to an army of more than 8bn robots, this might all smack of science-fiction. According to the IFR, even South Korean firms, by far the world’s keenest robot-adopters, employ ten manufacturing workers for every industrial robot—a long way from Mr Musk’s vision. In America, China, Europe and Japan the figure is 25-40 to one. The $25bn that, according to consultants at bcg, the world spent on industrial robots in 2020 was less than 1% of global capital expenditure (excluding the energy and mining sectors).
Yet as AI and robotics become more disruptive forces, workers will need something more than politicians promoting ahistorical fears. They’ll need policies to help workers increase their abilities to take advantage of change, not ones that acquiesce to fear of change, and present a vision of the future worth fighting for and living in.
In a new analysis, “Workers’ responses to the threat of automation,” researchers conducted an online experiment with 2,000 US workers who are randomly assigned to watch a video about automation or a placebo video. They measure their beliefs about automation risk, their employment responses (such as job search, training, and retirement plans), their preferences for taxation and redistribution, and their attitudes towards immigrants and trade. And what they find is that workers who receive information on expected automation change their preferences and attitudes, but not their employment responses. They do not intend to switch occupations or retrain, but rather join a union to protect their jobs. They also become less likely to vote and more left-leaning politically. From the analysis:
Taken together, the results from our survey experiment raise concerns about the potential for future automation waves to put pressure on public budgets through increased demand for redistribution and a larger welfare state. With democracy already on decline, we find that automation has the potential to stir anti-elite sentiments as well as reduce voter turnout and trust in politicians. Given the pending waves of automation, in particular in the service sector, policies need to carefully consider how to prepare workers for, and accommodate those left behind by, the changing landscape of work.
Do we have a political class ready for the challenge?
▶ New Room-Temperature Superconductor Offers Tantalizing Possibilities - Kenneth Chang, NYT | Scientists announced this week a tantalizing advance toward the dream of a material that could effortlessly convey electricity in everyday conditions. Such a breakthrough could transform almost any technology that uses electric energy, opening new possibilities for your phone, magnetically levitating trains and future fusion power plants. … The new superconductor consists of lutetium, a rare earth metal, and hydrogen with a little bit of nitrogen mixed in. It needs to be compressed to a pressure of 14,500 pounds per square inch before it gains its superconducting prowess. That is about 10 times the pressure that is exerted at the bottom of the ocean’s deepest trenches. … But it is also less than one one-hundredth of what the 2020 result required, which was akin to the crushing forces found several thousand miles deep within the Earth. That suggests that further investigations of the material could lead to a superconductor that works at ambient room temperatures and at the usual atmospheric pressure of 14.7 pounds per square inch. “This is the start of the new type of material that is useful for practical applications,” Ranga P. Dias, a professor of mechanical engineering and physics at the University of Rochester in New York, said to a room packed full of scientists on Tuesday at a meeting of the American Physical Society in Las Vegas.
▶ Improving Transparency in AI Language Models: A Holistic Evaluation - Rishi Bommasani, Daniel Zhang, Tony Lee, Percy Liang, Stanford University |
▶ Sam Altman invested $180 million into a company trying to delay death - Antonia Regalado, MIT Tech Review |
▶ Stanford researcher on the AI skills gap and the dangers of exponential innovation - Oliver Pickup, Raconteur |
▶ LLMs are not going to destroy the human race - Noah Smith, Substack |
▶ U.S. Shale Boom Shows Signs of Peaking as Big Oil Wells Disappear - Collin Eaton, WSJ |
▶ The Chatbots Are Here, and the Internet Industry Is in a Tizzy - Tripp Mickle, Cade Metz and Nico Grant, NYT |
▶ Working from home could be making it easier for couples to become parents—and for parents to have more children. - Derek Thompson, The Atlantic |
▶ It Turns Out Money Does Buy Happiness, At Least Up to $500,000 - Conrad Quilty-Harper
▶ How Google Became Cautious of AI and Gave Microsoft an Opening - Miles Kruppa and Sam Schechner |
▶ This geothermal startup showed its wells can be used like a giant underground battery - James Temple, MIT Tech Review |
▶ The US Air Force Is Moving Fast on AI-Piloted Fighter Jets - Tom Ward, Wired