Bill Gates and work

We are concerned that AI will cause massive job destruction. Influential and sensible voices—the latest being Bill Gates—tell us that perhaps this time it will be different and that when the shock of the technological transition stabilizes, we will see that enough jobs have not been created to replace the ones that disappeared. Below, I point out some of the factors that may influence this possibility.

1. AI could be the source of great benefits for humanity. It offers the prospect of much more productive economies (in net terms, which must take into account the need to minimize environmental costs) and of generating prosperity that the world, as a whole, will appreciate. Certainly, it would be good to slow down its progress to give us the opportunity to do it right, but unfortunately, it will not be possible. Neither the zero-sum competition of defense nor the positive-sum competition of international trade will allow it. If Europe exports to the world and a Chinese, or North American, competitor of a Catalan company lowers costs with the help of AI, our company will have to do the same, which indicates that the destruction of jobs in the sector of internationally tradable goods and services—that is, those that have the world as their market—is unstoppable.

2. A factor of great relevance is the evolution of ideas regarding to what extent personal income should depend on work. The historical association between income and work has already been diluting due to the secular trend towards the incorporation of income from capital returns (pensions can be seen in this light) and, above all, by subsidized public services, which do so in kind. It is highly desirable that this trend continues. Even the prospect of a universal basic income, within a highly productive economy, is not unthinkable. Personally, however, I do not believe it is desirable to completely break the coupling of "receiving from society" and "contributing to society". Both because I think it is better to build the social contract on this basis and because of the skepticism that AI will completely eliminate the role of incentives. In my view, the future will lead to an increase in income from capital (I will not comment on the magnitude) and to the fact that non-capital income will be organized around some simple principles: decent income for those who cannot work, a minimum income –subsistence– for those who do not want to work, a guarantee of decent income for everyone who works (which can be implemented, for example, through what is known as a negative income tax) and, above that, income derived from today's common labor contracts.

3. The question of whether there will be enough jobs can be approached from two angles: that of the normal development of the forces of supply and demand, and that of regulation. Supply will include the jobs that AI has not been able to destroy and the new ones generated as a result of AI. I dare not make predictions about it. Regarding demand, my opinion is that in economic sectors that are not internationally tradeable—that is, not export-oriented—there will be a significant demand for specifically human services. We will not want to be cared for only by robots. More likely, by humans with robots. Furthermore, to the extent that we live in prosperous economies, we will be able to afford, and perhaps prefer, to work 15 hours per week. Keynes anticipated this (in the same speech in which he introduced the notion of technological unemployment) a hundred years ago, although he expected it for our present day. Perhaps he was only wrong about the timing and we need to add a few more years. I hope not another hundred. Finally, it is evident that the demand for workers can be induced. For example, by providing (if we can afford it) that the working day be limited to three hours. Or, as Gates comments, perhaps it will be necessary to require that certain tasks be performed by humans. This can be done in a crude or subtle and moderately justified way. There is plenty of room for the needs of human supervisors. If we do this in the sectors that are not internationally tradeable and only in those (without therefore affecting the competitiveness of the economy very directly), I think it can be a reasonable compromise to avoid an economy where the income of a disproportionately large part of the population becomes disconnected from work.