Poker star Jimmy Chou, who has won more than $1 million playing the game, has a new teacher. Pluribus, a new artificial intelligence program, recently defeated Chou, along with a handful of the world’s best poker players, in six-player no-limit Texas hold’em. The strategy was computed in eight days, at a cost of $144 in cloud server power. As Chou graciously noted, “Whenever playing the bot, I feel like I pick up something new to incorporate into my game.” Playing against humans, Pluribus can win around $1,000 an hour, suggesting that online poker tournaments may soon become a thing of the past.
Six-player Texas hold’em now joins a long list of activities at which computers are superior to humans, including checkers (1995), chess (1997), Jeopardy! (2011), facial recognition (2014) and transcribing a telephone call (2016). It’s been two years since Google’s AlphaGo beat the world champion, Ke Jie, in the game of Go. The performance gap between AlphaGo and Ke is now about as large as the gap between Ke and a keen amateur. If Pluribus and AlphaGo were self-aware, they might look at our prowess in their games the way that we regard the intellectual powers of our pets.
Faced with rapid progress in artificial intelligence, factory automation and robotic logistics, some argue that that we should prepare for a world of widespread joblessness, embracing solutions such as a universal basic income. Others point out that predictions of technologically driven mass unemployment, including those of the Luddites in the 1810s and John Maynard Keynes in the 1930s, have been proven wrong.
Both the job-optimists and the job-pessimists are filled — to paraphrase Yeats — with passionate intensity. Yet we’re not as confident as either group. As the cliché goes, prediction is difficult — especially about the future. To ensure a successful future, a better approach for policy-makers is to follow the same strategy that families adopt when confronting risks such as a traffic accident or a home burglary: to take out appropriate insurance.
Let’s consider a few examples of what an insurance approach means for schools, colleges and universities, income security programs, and the private sector.
For individuals, there’s no better insurance policy against technological change than a terrific education. Yet over the past generation the academic aptitude of new teachers has declined. In reversing the trend, countries would do well to learn from Finland, a nation whose test scores were on par with those of the United States in the 1970s, but which has leapt ahead over the past generation by raising the status of teachers. In Finland, teaching is considered an elite occupation, and new teachers are drawn largely from the top tenth of the class. If any nation is effectively preparing its teenagers for a world of Pluribus and AlphaGo, it’s Finland.
Another education problem comes from the cost of post-secondary education. Five years ago, Joel and Eric Best called student loan debt in the US the “trillion-dollar problem.” Today, the total outstanding debt stands at $1.5 trillion. The cost and complexity of the system deter potential Albert Einsteins and Marie Curies from low-income and minority backgrounds. Simplifying the current financial assistance process would be a useful first step, but a smarter long-term strategy would be to shift from the current government-backed loan system to an Australian-style income-contingent loan scheme, in which repayments are made through the tax system, and only in years when graduates earn above the average wage.
A strong social safety net, another type of societal insurance, encourages productive risk-taking by innovators, since they need not fear falling into poverty if their business fails. We are also fans of the US earned income tax credit (EITC), which rewards work, is targeted at the most vulnerable and costs a whole lot less than a universal basic income. Enacted in 1975, the policy supplements earnings for low-income families. It has been proven to raise test scores among recipient children and cut the poverty rate. Any increase in the EITC would be desirable, but we particularly favour increasing the credit for childless workers and those with just one child. The EITC isn’t just good for children, it’s also a smart way of keeping people in employment. This expansion would provide a much-needed support for those with small families or without children.
Policy-makers should also insure against exploitative monopolists. Some markets are concentrated because the leading firms are better than the rest. But we ought to worry when monopolies exist because competitors cannot enter the market. Competition policy that lowers entry barriers and reduces switching costs is the best way of ensuring a level playing field for new entrants. For instance, we could achieve this in many digital networks by encouraging data and identity portability. When workers have more choices of where to work, customers have more choices about where to shop and everyone has more choices about their social networks, it cuts down the risk that comes from putting all our eggs in the same basket.
Artificial intelligence is both daunting and exciting. We can’t be sure what the future will hold, but if we limit the downside and foster the upside, we can raise the odds of a more prosperous and egalitarian future. No poker player — not Jimmy Chou, not even Pluribus — can be sure what cards they will be dealt. But with the right policies in place, society can increase the chance of a winning outcome.
Do you have something to say about the article you just read? Be part of the Policy Options discussion, and send in your own submission. Here is a link on how to do it. | Souhaitez-vous réagir à cet article ? Joignez-vous aux débats d’Options politiques et soumettez-nous votre texte en suivant ces directives.