Part 3 of Roger Bootle’s The AI Economy is called “What Is to Be Done?” Here he addresses some of the policy questions posed by the advances in artificial intelligence and robotics.
Since Bootle does not accept the most pessimistic or apocalyptic visions of the technological future, he does not see any good reason for government to discourage or impede the adoption of these new technologies. He does not support a “robot tax,” based on the assumption that robots will deprive humans of the opportunity to work or deprive government of needed revenue from employment and income taxes. He expects new forms of work to become available, and incomes and tax revenues to rise along with enhanced productivity.
However, he does think that new regulations will be needed to protect the public from some of the misuses of technology. For the most part, these are problems that are already in the spotlight, such as cybercrime, massive collection and storage of personal data, and the increasing surveillance of populations by authoritarian governments.
A very tricky question is who will be legally responsible for the decisions made by AI systems. The algorithms they use are so complex that determining how a particular decision was reached can be extremely difficult. But someone must take responsibility for decisions that impact people’s lives. “The algorithm made us do it” cannot be a sufficient excuse. Social media companies are grappling with the question of whether their algorithms contribute to political radicalization by feeding their customers more and more of the political messages they “like,” no matter how inaccurate, hateful or defamatory. As I write this, Elon Musk has just bought Twitter and vowed to weaken its rules about acceptable content. (The self-appointed “Chief Twit” has also set a despicable example by tweeting unfounded conspiracy theories about the assault on Nancy Pelosi’s husband.)
Bootle imagines a transformation of education, but what he expects is not what the most ardent admirers of artificial intelligence often envision:
AI enthusiasts gush about the way robots and AI are going to transform a particular aspect of society—in this case, education—but they completely misjudge the nature of that transformation. Supposedly, teachers are bound to go the way of taxi and lorry drivers, set to be made redundant and consigned to become yet another lump of unemployed skilled workers having to scrabble around for some new way of earning a living. Meanwhile, the subjects that need to be taught are revolutionized, with traditional arts subjects thrown out of the window, to be replaced by STEM subjects only.
Bootle fears an overemphasis on science, technology, engineering and math, to the exclusion of a broader education. This is consistent with his view of humans as creative thinkers and decision makers, assisted by technology but not replaced by it. He quotes David Kosbie of Carnegie Mellon, who says that “with AI taking over routine information and manual tasks in the workplace, we need additional emphasis on qualities that differentiate human workers from AI—creativity, adaptability and interpersonal skills.” Bootle’s view is also consistent with his expectation that workers will take some of the benefits of higher productivity in the form of more leisure. People need education for life, not just for technical proficiency. He agrees with those who would add an A for Arts to make STEM into STEAM.
This does not mean that the organization and methods of formal education will survive intact. The traditional lecture method may well be replaced by more efficient methods of imparting information, but that doesn’t mean that most teachers will have nothing left to do. As with other creative forms of work, teaching can be technologically assisted without being technologically replaced. If teachers are to work more creatively with students, we may actually need more teachers, but with smaller classes.
Bootle describes the higher educational system as an “expensive disgrace.” Too many students pay too much to receive a degree that gives them an advantage over other job applicants without necessarily enabling them to do a better job. Here he draws on the work of other critics of “credentialism,” such as economist Bryan Caplan (The Case against Education). Bootle admires Switzerland, which has one of Europe’s leading universities in Zurich, but also more vocationally relevant alternatives for young people who don’t go to university at all. He believes that “the bulk of educational time can be spent at home or in the work place, interspersed with occasional visits to the physical ‘seat of learning.'” Higher learning need not be packed into four years of frenetic activity, but should be spread over a lifetime.
One issue on which Bootle and Caplan disagree is public education. Caplan would separate school and state, leaving education to the private sphere. Bootle uses the economic concept of “externalities” to argue for a large public role. The costs and benefits of an economic exchange are not limited to the individuals who engage in it. When a company pollutes the environment in order to produce a product, the costs are borne by society. Similarly, when an educator well serves a student, the benefits are shared with society. A democratic society has good reason to incentivize activities with positive externalities and disincentivize those with negative externalities. If left to their own devices, private firms have a limited incentive to invest in their workers’ lifetime learning or retraining, since the firm doesn’t own the workers for life and recover the full cost of its investment. The need for more education in the technological era will probably increase the government’s role in financing education.
Bootle expects artificial intelligence and robotics to cause some disruptions to employment, but he does not think they will be severe enough to require drastic policy measures.
There is no reason to believe that the Robot Age implies the Death of Work. There will be plenty of jobs for people to do… Even so, it is possible that the jobs that are available for many people are temporary, insecure, and badly paid.
The idea of a Universal Basic Income (UBI) provided by government has been kicking around for many years, but the possibility that technological change may destroy many jobs has sparked new interest in it. If Depression-level unemployment were to become the norm, the economy might need a UBI to sustain aggregate demand. Although that sounds like a liberal idea, some conservatives like it because it may enable firms to pay lower wages, and also simplify the government bureaucracy that administers the complicated system of benefits we have now. Bootle does not expect the level of unemployment that would require a UBI; nor does he consider it very workable. If the UBI is truly universal and set relatively high, it may spend too much tax revenue on people who don’t need it. If it is set too low, it provides little help to the people who need it most. Bootle agrees with economist John Kay, who said, “Either the level of basic income is unacceptably low, or the cost of providing it is unexceptionally high.” In 2016, Switzerland defeated a UBI proposal in a referendum. In 2017, Finland experimented on it with a small sample, but decided against extending it.
I was surprised that Bootle discussed UBI at great length, but rejected the idea of a Basic Jobs Guarantee (BJG) out of hand. From an economic standpoint, I would think that paying the unemployed to do some useful work, while hopefully developing their job skills, would be more cost-effective than paying them to remain unemployed. Modern Monetary theorists advocate BJG as an economic stabilizer, since the program would expand in times of economic downturn, but contract in times of fuller employment. Bootle observes that it has not generated much support yet on either side of the political divide.
What can be done for workers and families who are struggling in the emerging economy? Bootle points out that even if economic inequality is here to stay, new technologies can raise the average income, benefitting people at all levels. He also hopes that new technologies will help “improve the efficiency of public services, and hence deliver better results for the same money, which would disproportionately benefit the less well-off.” But he repeatedly returns to the issue of education, where he believes that public investments have the best hope of reducing inequality.
At the end of the book, Bootle restates his main conclusions:
- The technological breakthroughs of our era will have consequences comparable to those of earlier phases of the Industrial Revolution. They will raise productivity and the return on capital, destroy some jobs but create others, and generally provide higher incomes and more leisure.
- Smarter machines will do many things, but humans will still do some things better or at lower cost.
- The AI revolution will liberate people from jobs that have “taxed their spirit and eroded their strength and enthusiasm,” leaving them with more freedom to be fully human.
- The best thing the state can do to help is to fund public education, with a greater emphasis on lifetime learning and retraining.
Many readers who are familiar with the more apocalyptic visions of science fiction writers and other technological visionaries may wonder how Bootle can be so optimistic. How do we know that smart machines won’t replace humans in more dramatic ways than he envisions? Bootle realizes that in the end, defining the difference between a human and a machine requires a deep dive into philosophy and science, opening up fundamental questions that lie beyond the scope of his economic perspective. He wisely leaves these questions for his Epilogue, and even there only hints cautiously at the ultimate answers. I will comment on that section in my last post.