You may need approximately 6 minutes to finish reading this article, which contains 633 words in total.
How to spot a genius (1)
Ervin Macic was despondent. While in school he twice won medals at the International Mathematical Olympiad and researched artificial intelligence, trying to speed up how models make predictions. He dreamed of one day joining an AI lab to make the technology safe. Yet the 19-year-old Bosnian prodigy was unable to take a spot at the University of Oxford: its fees of £60,000 a year were five times his family’s annual income. So he went to the University of Sarajevo, where he sat programming exams on a decades-old IBM computer.
Mr Macic’s case is far from unique. Around the world vast amounts of talent goes to waste. Economists speak of “lost Einsteins” who might have produced transformative work had they been identified and nurtured. Nowhere are the consequences clearer than in AI, where the scarcity of top researchers allows a tiny cadre to command CEO-level pay. Governments that lavish billions on semiconductors to win the AI race neglect the talent that drives progress. Brains, treated with the same urgency, may prove a better longer-term investment. What might an industrial policy for talent look like?
For now such policy amounts to procurement, not production. Governments focus on the last step: enticing existing superstars. The contest is fiercest between China and America. China’s Thousand Talents Plan, set up in 2008, aims to lure back citizens trained at elite foreign programmes; this October it will add a flexible K-visa to attract STEM specialists. America counters with the O-1A visa and EB-1A green card, both reserved for individuals of “extraordinary ability”. Other countries dabble. Japan has announced a $700m package to recruit top researchers. The EU’s “Choose Europe” scheme promises to make it a “magnet for researchers”.
A more extreme scarcity mindset about superstar talent drives the scramble among firms—and helps explain the premium now placed on brains. As they race to build ever-larger models, individual researchers are seen as capable of unlocking breakthroughs worth billions. Sam Altman, OpenAI’s boss, once quipped about “10,000x engineer/researchers”, ultra-productive coders whose output can transform a field; the idea has since become industry lore. Elite researchers command valuations once reserved for companies.
These bidding wars rest on two assumptions. One is that a few superstars make outsize contributions; the other is that the supply of such talent is fixed. The first assumption is well founded. Breakthroughs are produced by a small elite: the top 1% of researchers generate over a fifth of citations. James Watt’s refinements to the steam engine helped launch the Industrial Revolution. More recently, Katalin Karikó’s lonely pursuit of mRNA technology paved the way for covid-19 vaccines. Individuals can shift the frontier for everyone.
The second assumption is shakier, however, for much potential never flowers. Geography is the first barrier. Some 90% of the world’s young live in developing countries, yet Nobel prizes overwhelmingly go to America, Europe and Japan. According to Paul Novosad of Dartmouth College and co-authors, the average laureate is born in the 95th percentile of global income. Although some disparity is to be expected, the scale suggests much talent does not have a chance to flourish. Similarly, Alex Bell of Georgia State University and co-authors find that American children from the richest 1% of households are ten times more likely to become inventors than those from below-median incomes. They estimate that closing America’s class, gender and race gaps in invention would quadruple the number of innovators, sharply raising the pace of discovery.
Where should governments begin their search for genius? One tempting answer is at the very top of the funnel, increasing the number of children who ever have a chance to develop their abilities. Universal fixes—improved nutrition, better schools, safer neighbourhoods—could help. But the problem is that, given how rare genius is even when better identified, such schemes are by their nature poorly targeted.