Here are two different versions of American nationalism:
Version 1: High tech is evil, part of the woke conspiracy to radicalize America. We need to go back to muscular old industries like coal and steel and autos, which employ lots of blue color workers. We need to revive the Rust Belt. EVs are a fad. Brick and mortar stores are better than Amazon. Silicon Valley is full of lots of immigrants from non-white countries. The Bay Area no longer looks like America.
Version 2: China is our greatest adversary. The battle of the 21st century will be determined by who controls AI, which will impact the battle for military supremacy. We need to import vastly more high skilled engineers from places like India and China. This will strengthen America and weaken China. Greater San Francisco is our most important city, by far. It has the smartest people and it is where the future of AI will be determined.
I am certainly no expert on artificial intelligence. But people much smarter than me insist that military supremacy in the 21st century will be determined by which country achieves AGI first.
It seems to me that there are two types of American nationalism. One type might be called “nostalgic nationalism.” A yearning to recreate the America of the 1950s. Another type might be called “future-oriented nationalism”, and focuses on winning the race to control AI and thus dominate the world.
I’m not at all sure that what I’ve called future-oriented nationalism is actually nationalism at all. In that case, it might make more sense to think of a single nationalism with internal contradictions. Nostalgic nationalism is the real thing, but it’s full of internal contradictions. Its proponents wish to go back to the 1950s, but they also want to confront the threat of a rising China. Can we do both?
PS. Neither of these versions of nationalism reflects my own view.
READER COMMENTS
Majid Hosseini
Jun 24 2024 at 12:27am
This is pure nitpicking, but it is mistake that gets repeated so often that it is worth pointing it out. It is also important in some ways as far as the culture of Silicon Valley is concerned. The center of Silicon Valley isn’t San Francisco at all and has never been. It’s Palo Alto and the cities to its south (all in Santa Clara valley, hence the name Silicon Valley). Nvidia, AMD (and Intel) are in Santa Clara. Google is in Mountain View. Apple is in Cupertino. Facebook is in Menlo Park, bordering Palo Alto to its north.
In recent years, as the industry shifted from hardware to software, there was some movement by a few tech companies to either start from San Francisco (Twitter and Uber) or open offices there (Google and Facebook), mostly because a sizable number of tech employees these days (unlike the earlier days of Silicon Valley) consists of younger unmarried people who wants to live in a more urban place with more entertainment, unlike the suburban lifestyle of Silicon Valley. The city also during the tenure of the late Mayor Ed Lee gave tax breaks for opening offices in the Tenderloin area which made it attractive. But San Francisco never was the center. All of major VC firms are on Sand Hill Road, at the boundary of Palo Alto and Menlo Park, and as I mentioned before all major companies are located in Silicon Valley. And after Covid, many companies closed their San Francisco office.
Scott Sumner
Jun 24 2024 at 1:46am
“Greater San Francisco is our most important city”
The term “greater” refers to the entire metro area, not just SF.
Majid Hosseini
Jun 24 2024 at 2:37am
It is clear what you’re referring to and my comment has no importance for someone outside of the Bay Area (and probably neither for anyone living in the Bay Area). But from commuting, living, culture, etc perspective, they’re very separate places. They’re also in different MSAs according to BEA. Personally I think Silicon Valley is ambiguous enough these days that it is perhaps the best term (people tend to use it for Santa Clara + San Mateo + San Francisco counties together, even though it’s historically and geographically only Santa Clara county)
Henri Hein
Jun 24 2024 at 1:52pm
I agree with your point, Majid. I worked in Sillicon Valley for 20 years. San Francisco and Sillicon Valley mentally at least feels different. The subway line doesn’t even go to Palo Alto (it barely reaches San Mateo). I wonder if it’s a bit like people in Anaheim insisting they don’t live in Los Angeles. We didn’t have any animosity towards SF, but when people would say I lived or worked in SF, I would respond, no, I really don’t, but I can understand why you might see it that way.
But it is indeed a nitpick.
Majid Hosseini
Jun 24 2024 at 12:57am
On AGI: I don’t know about the entire century, but I don’t think we’ll see it in our lifetimes. There’s absolutely no path to AGI with large language models, and I’m not sure anyone has any concrete ideas that could work. Yan LeCun is the only prominent person that has articulated some ideas for moving forward, but I believe they’re purely speculative in nature right now. I do see some specific-use systems getting built with a separate logic engine in tandem with an LLM, but that’s not AGI
Scott Sumner
Jun 24 2024 at 1:47am
I don’t have an opinion, but some experts think we are only 3 years away from AGI.
Kenneth Duda
Jun 24 2024 at 2:10am
This is hopeless to predict. AI experts were polled in 2014 when computers would beat the best human go players, and predicted 10 years. Actual answer was 10 months. Who knows when AGI might be achieved but given the recent rate of progress, counting on coal and steel for military supremacy doesn’t look like a very smart bet.
Majid Hosseini
Jun 24 2024 at 2:44am
I agree it’s impossible to predict, but I think it’s a fairly safe bet to say LLMs won’t get us to AGI and there’s right now no clear alternative path forward. LLMs, like almost everything else in machine learning, at core are a form of regression. I don’t think we can get to human level intelligence with any regression-like models. It’s possible someone comes up with something entirely new, but as you mentioned, it’s impossible to predict that.
Scott Sumner
Jun 24 2024 at 2:00pm
Given that we have no idea what human intelligence is (perhaps it’s also just “regression”), I am reluctant to make predictions in this area.
Mactoul
Jun 24 2024 at 2:30am
America didn’t need the Chinese or Indian engineers to reach Moon Or to invent nuclear bombs.
Why should America need them now, with twice the population.
Jon Murphy
Jun 24 2024 at 7:24am
America has long been home and fueled by immigrant brilliance. In fact, the two programs you mentioned heavily involved immigrant scientists: the Manhattan project relied on many immigrants from Europe, Russia, and Asia (check out the roster at some point. It’s a Who’s Who of the best physics minds in the world). NASA was famously staffed with German scientists.
Immigration has always been what’s made America great and unique in the world. Our willingness to embrace good things, regardless of where in the world they started, has led us to be an epicenter of technology, research, culture, and wealth.
Allow me to answer your question with another question: why shouldn’t we use brilliant minds from other countries?
MarkW
Jun 24 2024 at 8:30am
No, not Indian and Chinese scientists back then — but does the name Wernher Van Braun mean anything to you?
MarkW
Jun 24 2024 at 8:24am
Neither of these are my views either, but each have some reasonable elements. High tech has been woke (though less now than it was a few years ago) and has done some evil things (it didn’t take much arm-twisting for them to get onboard with the whole censorship-by-proxy effort to suppress dissenting views). Republicans were not wrong in seeing big tech companies as powerful political opponents. But at the same time, high-tech is our most important growth industry, and AI is going to have a huge economic impact regardless of military uses. Silicon Valley IS full of immigrant engineers (and the Bay Area doesn’t look like a lot of the US) — but that is a good thing that should be encouraged. EVs aren’t a fad, but proponents have gotten way out over their skis. It has been a mistake to try to force a rapid transition through subsidies, mandates (and tariffs!?). In the near term, there is going to be a lot more (expensive) pain in the EV industry as a result of these mistakes.
Scott Sumner
Jun 24 2024 at 2:03pm
“It has been a mistake to try to force a rapid transition through subsidies, mandates (and tariffs!?). ”
I agree. But note that our tariff policy is actually dramatically slowing their adoption (by even more than the subsidies boost EV adoption.)
MarkW
Jun 24 2024 at 6:14pm
Yep. But the EV proponents pretty much have to do the tariffs for political reasons. They’ve been trying to sell to voters on a glorious green energy/green jobs future, and the whole vision falls apart if those green EV auto-making jobs are mostly offshore while the ICE auto-making jobs they replace are mostly domestic. The same logic applies to solar panels.
steve
Jun 24 2024 at 9:10am
Winning the AI battle might mean an economic victory of sorts but I doubt it leads to a military victory. I also think that the Chinese economy is still so centrally controlled that winning the AI battle wont give them the big economic win some fear. I worry a bit more about the second scenario. Looking backwards doesnt move an economy forwards. Even if we do manage to regain the lead in auto production it will mean we greatly increased productivity cutting out lots of jobs. Telling people you are bringing back all of those jobs is good election politics as people want to believe it. It’s not a good economic plan.
Steve
Peter
Jun 24 2024 at 4:36pm
AGI?
Thomas L Hutcheson
Jun 25 2024 at 8:00am
I’m closer to the techno-nationalism strand but see it mainly just adding urgency to all the “growth with equity” things we should be doing even if China were not a geopolitical rival: freer trade, merit based immigration and lost of it, taxing net CO2 emissions, consumption taxes to close the deficit, land use and building code reforms to permit higher value housing nan commercial real estate development, reforming regulator agencies to use cost benefit analysis in their decisions, all the standard neoliberal stuff.
Jeff
Jun 25 2024 at 10:15am
In your linked piece, the writer says that the current AI research trajectory is up against a fundamental wall but that really smart people are working hard at it and “they will probably figure out some crazy stuff.” So, in essence, it is an science project that will need future undetermined breakthroughs to happen precisely on schedule in order to stay on the many trend lines the author extrapolates well into the future.
I wasn’t surprised when I learned that the author’s fundamental training was in economics. Forgive my impertinence but I feel his piece betrays some of the flaws I find common in economists’ thinking. While they tend to have excellent mathematical chops, they are rarely humbled by the study of those pedestrian objects of the natural world that actually do seem to be constrained by mathematical laws. So they can produce what look like highly quantitative treatises on sexy concepts like “intelligence” or “growth”, but, upon pulling back the curtain, it becomes clear that their empirical analyses lean heavily on the drawing of trend lines through highly imperfect surrogate measures.
Scott Sumner
Jun 25 2024 at 3:22pm
My impression is that most economists are much more skeptical of AI than are most people in the AI field.