When you’re young, there’s a fine line between wonder and confusion. For me, nowhere was that clearer than our living room’s 8-track tape player. The ability to instantly shift – mid-song – from track to track of my mother’s beloved Sound of Music 8-track cassette led to wonder at mystical musical connections between parts of Sixteen Going on Seventeen (track 3) and The Lonely Goatherd (track 4) serendipitously recorded on the same bit of ribbon, and confusion at who would invent such bizarre technology.
Equally confusing was another favorite 8-track: Elton John’s Goodbye Yellow Brick Road. I spent hours trying to figure out what Candle in the Wind (track 2) had to say about Saturday Night’s Alright for Fighting (track 3). But even more puzzling was the cover. While harder to make out than on the much larger album, it depicted the artist – resplendent in a pink satin jacket with Elton John in large green letters on the back, white bell bottoms, and sparkly ruby platform shoes – stepping through a movie poster and firmly onto the eponymous yellow brick road. Quite the opposite of saying goodbye to the yellow brick road, Elton was clearly saying hello.
In retrospect, this was a mistake that would have been caught had the good people at the record company not been doing so many drugs. Today, as digital technology has left 8-track tapes in the yellow brick dust, even the most irresponsibly Dionysian label can avoid confusing youngsters via the use of artificial intelligence (AI). AI is no longer sci-fi. It’s already helping identify cancer cells, detect incorrectly wired bolts in jet engines, find financial fraud, and correctly identify pastries. It can certainly figure out whether a recording artist is coming or going without being mesmerized by ruby shoes or yellow roads.
AI is shorthand for a system of connected digital technologies that result in improved decision making and better products and services. There are three primary components of AI: (1) a series of algorithms, each charged with a specific task e.g., determining whether a sign is octagonal; (2) a very large set of relevant training data that helps the algorithms learn (“machine learning,” but beware bad data; if every photo of a bedroom shows a neatly made bed, AI will have a hard time classifying a normal bedroom); and (3) layers of connections that shift to optimize correct outcomes, akin to the way neural networks form in the human brain (machine learning becomes “deep learning”).
AI isn’t a single thing, just like technology isn’t a thing. A better analogy is electricity, which Thomas Edison famously described as the “field of fields… [holding] the secrets which will reorganize the life of the world.” AI was already growing like topsy before Covid, reaching 37% of organizations. But labor limitations over the past year have resulted in incredible increases. KPMG reports 37% penetration growth in financial services, 29% in retail, and 20% in tech. In a new Deloitte survey, two-thirds of business leaders already say AI is critical to remaining competitive, and 57% predict AI will “substantially transform” their company within three years.
If the first and second industrial revolutions supplemented labor with capital, the third is supplementing labor with software. AI is the neural root of this new revolution. And while I remain deeply concerned about the future of work – and the dialectic between lost jobs and freeing up remaining workers to be more creative (a debate that’s keeping Future of Work pundits in pastries) – I’m even more concerned about the future of our country.
Last month, a National Security Commission on AI led by former Google CEO Eric Schmidt, and including leaders of some of America’s largest tech companies, released a lengthy report that concluded “America is not prepared to defend or compete in the AI era.” Citing not only economic competitiveness, but also threats like AI-powered disinformation attacks, cyber attacks, and – gulp – military technology, the report frames the AI competition with China as akin to the Cold War with the Soviet Union. China’s stated goal is to surpass the U.S. as the world’s AI leader in this decade and to dominate by 2049, the 100th anniversary of the People’s Republic. The report concludes China is already “an AI peer in many areas and an AI leader in some applications” and warns that China’s internal use of AI as a tool of repression is “a chilling precedent… [and] a powerful counterpoint to how… AI should be used.” (As Uyghurs and Hong Kong democracy protesters have learned, lack of individual freedom + AI = dystopia.) The 2020s will be as critical for this competition as 1945-55 was for the Cold War, and perhaps more as AI technology tends to build on itself.
The only way to ensure the AI war never turns hot with China or any other authoritarian regime– to keep looping Candle in the Wind rather than Saturday Night’s Alright for Fighting – is to sprint far ahead. The key to this is talent. One recent estimate pegged the number of people globally with the skills to create fully-functional machine learning systems at under 10,000. A second report estimated between 22,000 and 37,000. And while America is currently ahead (46% working for a U.S.-based employer vs. 11% for China-based employer), a more refined estimate situates 29% of top-tier deep learning talent in China. Regardless, given the scale – breathtakingly small given the imponderably high stakes – it wouldn’t take much to tilt the playing field. On this point, the Commission bangs its shoe on the table: “the human talent deficit is… [America’s] most conspicuous AI deficit.”
In a country where the number of U.S.-born students participating in AI doctoral programs has remained flat for 30 years, we need to produce more homegrown talent. Colleges and universities should be making it much easier and attractive for students to opt in to STEM – for example, by transforming STEM courses from lectures to active learning, eliminating unnecessary and weed out courses, and establishing many more applied STEM programs – and harder for students to opt out of STEM entirely. Each and every college graduate must be equipped with core data and digital skills.
With respect to AI-specific training, while new last-mile training programs will play an increasingly important role, colleges and universities should be launching dozens of new data and AI-oriented majors, concentrations, and certificates. New graduate programs should be a top priority. The Commission on AI recommends a second National Defense Education Act (the first was in 1958) including “thousands of fellowships in fields critical to the AI future.” At the same time, not all AI pathways lead to terminal degrees. Many entry-level AI-oriented jobs require only statistics, data skills, business intelligence technologies like SQL, and basic knowledge to apply these skills to a specific industry or function. For colleges struggling to recruit students increasingly focused on employment outcomes, there’s no better way to address concerns that jobs will be lost to AI than by massively scaling up efforts to prepare students for AI jobs.
As AI is entirely dependent on more and better data, open systems produce stronger AI than closed systems; AI is a reflection of the society that produces it. So as we open up from Covid, we must also open up for AI. This means proactively recruiting AI talent from around the world. In recent years, the U.S. hasn’t seen nearly as much AI talent inflow per capita as Canada or the UK, while China has significantly bolstered its ability to attract researchers who earned Ph.Ds in other countries. Currently, China has just as many AI researchers who trained in the U.S. as vice versa. The Commission recommends a comprehensive immigration strategy for AI talent, including stapling Green Cards to AI-relevant STEM degrees.
Colleges and universities should lobby hard for this, and for the Biden Administration to match the smart marketing of Canada and the UK by allowing F-1 visa international students who graduate in any STEM field to work for at least one year without a separate OPT application. After four years of closing the door to international students and watching international enrollment decline, including a 70% drop in Chinese students this year, if patriotism isn’t sufficient cause to exercise their collective clout, colleges and universities should consider how incremental tuition revenue from international students could plug projected holes in operating budgets. Because despite Trump’s best efforts, Chinese families still view U.S. universities as the gold standard in higher education.
Another favorite 8-track cassette in our house was Tom Lehrer’s That Was The Year That Was. Lehrer – a mathematician who worked for the NSA before teaching at MIT – was a satirical musical genius. This recording of a live performance tackled the political, religious, environmental, and educational controversies of 1965 in musical styles ranging from folk to ragtime, and provided this Canadian Gen Xer with a somewhat skewed view of Cold War America.
The most pointed song on the record was Wernher von Braun, an ode to the Nazi rocket scientist who, after being captured by the U.S., led the development of ballistic missiles and the Saturn V rocket that took Americans to the moon:
Gather 'round while I sing you of Wernher von Braun
A man whose allegiance
Is ruled by expedience
Call him a Nazi, he won't even frown
"Nazi, Schmazi!" says Wernher von Braun.
Don't say that he's hypocritical
Say rather that he's apolitical
"Once the rockets are up, who cares where they come down?
That's not my department!" says Wernher von Braun.
Some have harsh words for this man of renown
But some think our attitude
Should be one of gratitude
Like the widows and cripples in old London town
Who owe their large pensions to Wernher von Braun.
The brilliance of Lehrer’s song is that it satirized its subject while revealing just how hard-hearted America had become about the need for talent at the height of the Cold War. We were on war footing and managed to ignore a great deal.
The battle for talent will be decisive in the coming AI war and winning it will require a much higher level of common purpose and urgency from major players in the talent economy. Colleges and universities – reputedly able to see further and clearer than most civil institutions – should be sounding the alarm, particularly because this competition is less likely occur through headline-worthy blockades and proxy wars than away from the spotlight, in company and research lab innovations, but nonetheless will have profound effects on the way we live within a short period of time. But American higher education has not yet recognized the scale and immediacy of the problem. In stark contrast to the Cold War, you wouldn’t know it from looking around.
A hard-hearted common purpose on AI talent also means reversing a raft of shortsighted, tit-for-tat Trump Administration policies, including lifting restrictions on visas for members of the Chinese Communist Party, restarting Fulbright and Peace Corps programs in China, and allowing China to run Confucius Institutes on college and university campuses. We can’t allow fear to close off America. Talent will win the AI war, and the AI talent that matters wants open, not closed. As the AI Commission report suggests, counter-espionage efforts can be highly targeted and shouldn’t close off the country.
The good news is that if higher education can help America to not only open, but aggressively open – to all, and to Chinese scientists in particular (regardless of whether they were members of the Communist Party) – we’ll force authoritarian regimes like China to close further. The more we’re open and appealing to AI talent, the greater the AI brain drain from China, the more China will have to close, which will further slow its AI capabilities. Aggressively opening will also undermine the narrative of a dysfunctional U.S. that Xi Jinping has propagated over the past decade. The only way that narrative persists is if we stay closed or only open halfheartedly. The past year has given us a head start down this road; due to Covid, China closed its economy more than any other major country.
The battle over AI talent also provides an opportunity for higher education to win back public affection. Over the past decade, politicization of higher education combined with the deepening crises of affordability and employability caused American higher education to lose public support. But just as the Cold War mobilized an historically individualistic population around a reasonably compelling goal of national survival, colleges and universities can regain lost goodwill by assuming the mantle of technological and economic competitiveness in an AI era, and perhaps defending democracy against autocracy. And don’t be surprised if regained goodwill translates into dollars – not only in terms of increased enrollment, but also in government support; federal R&D funding for universities grew nearly 7x between 1953 and 1967. And while policymakers may debate whether the government should be making billion-dollar bets on specific technologies, there’s little disagreement about the wisdom of investing in talent.
Tom Lehrer was farsighted. To wit, the last verse of Wernher von Braun from all the way back in 1965:
You too may be a big hero
Once you've learned to count backwards to zero
"In German, oder Englisch, I know how to count down
Und I'm learning Chinese!" says Wernher von Braun.
While America has many skill gaps, only one is truly existential. AI will produce a world of wonder. But there should be no confusion: producing and attracting more AI talent will be essential to disproving autocratic China and Russia’s premise that – in the words of President Biden – “autocracy is the wave of the future and democracy can’t function.” American colleges and universities must lead the way so the Wernher von Brauns of AI won’t ever have to learn Chinese.