Future of Professions: How Artificial Intelligence Will Affect the Labour Market and Inequality
The development of artificial intelligence (AI) technologies is rapidly gaining momentum, and economic consequences of its emergence are drawing more and more attention of researchers, politics and regulatory authorities. The economic policy in this field brings in focus not only prerequisites and stimulation of the AI technology expansion but also their consequences for the economy. Those consequences are most likely to be the most substantial for the labour market and the structure of the socioeconomic inequality.
Will a Robot Take Your Place?
In the opinion of Randall Collins, a social scientist, we are witnessing now another round of technological substitution, during which machines will take place of employees of “obsolete” professions. This process does not come to a simple improvement in the labour effectiveness and savings on costs. It is preparing for us at least two serious challenges.
Challenge number one: The algorithms have already successfully accomplished a part of tasks of financial traders, investment bankers, health care professionals, and journalists. That is to say that, unlike the industrial revolutions of the past, today’s technological substitution hits, first of all, the middle-class professions that require specialized education.
Challenge number two: Despite the fact that technological substitution is accompanied by the process of creating new profession niches and jobs, there are no guarantees that the number of new jobs will increase at the same rate as technological unemployment.
Economists look to the future little more optimistic, but they also predict labour skills and technologies mismatch in the short term. In the distant future, AI will replace only part of the tasks carried out by people today, while new niches in the labour market will emerge in areas where human abilities have a comparative advantage. Each breakthrough in the data acquisition and processing technologies more clearly outlines people competencies that cannot be delegated to a machine: an algorithm analyzes the data and makes forecasts, but nevertheless, decisions are taken by people. There is a consensus in opinion of most researchers that the implementation of AI will make much-in-demand the “human” capacity in strategic management, emotional and cognitive labour, management of human relations in organizations, teaching of other people and so forth.
There exists an opinion that technological substitution in the areas where the implementation of AI will be the most efficient would go faster, and the share of these branches in the national GDP will decline. Through this process, the sectors of the economy which demonstrate the most dynamic growth in the number of jobs would have probably nothing to do with the use of AI and other automation technologies, though it is AI and other automation technologies that a critical mass of new jobs would accrue to; and in the circumstances a data analyst will not be a “profession of the future” at all.
Theoretical economics supposes four case scenarios, which could be realized in various combinations.
Scenario number one: Innovative technologies may lead to the full automation of human labour in certain industries and, consequently, to the extinction of related professions (for example, delivery men or call-centre employees.)
Scenario number two: Technological substitution can affect only specific working practices and particular tasks, leaving space also for man’s work (this scenario appears to be likely for health care and legal proceedings.)
The Economic Gap Between People?
Although theoretical predictions are optimistic in the long run, the immediate effects of technological substitution on the structure of social inequality require some serious attention. According to research of 2016 conducted by the Executive Office of the President of the United States, the median probability of full automation is 83% for low-wage jobs (median hourly wage of $20 and less for 2010), while it goes down to 31% for middle-wage jobs ($20–40). Last, the technological substitution of types of jobs being paid $40 per hour and more (the median value of 2010) is only 4%. As the level of remuneration often depends on the educational grounding, the above statistics mean that the beneficiaries of the AI expansion will be, first of all, highly-educated people holding well-remunerated jobs. Low-paid employees with a low qualification will lose their sources of means of subsistence. Moreover, since the education system is also one of the major drivers of the inequality of income, this inequality may be exacerbated: if the trainability of new skills correlates with education acquired, but not its direct consequence, simply increasing access to education could not remedy the situation.
Progressive increase of the share of capital in the economy is another effect of AI that can aggravate inequalities between people. According to Thomas Pikkety, where the return on capital is greater than the rate of economic growth, this circumstance leads to an increase in economic inequalities. If AI is the new and high-efficient form of capital, its expansion would exacerbate the highlighted tendency.
An alternative scenario supposes that artificial-intelligence technologies will cause a new wave of de-skilling which, as Collins predicts, will affect highly paid types of jobs and middle-class professions in the first place.
For example, one of the most important tasks in medicine is diagnostics which conceptually is a kind of prognostication. In that way, the application of algorithmic diagnosis which by definition is cheaper and probably more reliable will replace the work of medical professionals. On the other hand, the application of advanced technologies in replacement of the work of sick attendants, nursing staff, etc. does not make economic sense. Nevertheless, historical experience says that more qualified labour better adapts to technological substitution, therefore, this scenario is less likely.
Expert opinions regarding the economic consequences of AI are not much different from similar discussions about the new technologies of the past. True innovations by definition disturb the existing balance and at the same time offer significant advantages in the medium-to-long term. It is important to realize, however, that the need to adapt the existing institutions, i.e. the labour market, the education system, social protection measures and regulation, proceeds from not only short-term problems caused by the development of AI: in the final count, a guarantee that all benefits promised by the new technology be realized is an institutional and regulatory “response” to the challenge posed by artificial intelligence. That is why it already now makes sense to pay close attention not only to the rate of registration of patents or the growth rate of investments in that sphere but to related proposals for reforming social and economic policies: from universal basic income to the taxation of robots, no matter how surprising they seem.