The encroachment of automation and robotics in the workplace has forced us to rethink the way some work is done, and this has raised concern as to whether there will be enough jobs at the future for human workers who need it. Much of the attention so far has focused on blue-collar workers, as factory assembly lines and warehouses have adopted automated processes faster and more visibly than other industries. Automation in a workshop evokes a simple image: robotic arms assembling parts in Tesla cars; mobile robots driving pallets of goods through Amazon distribution centers. In both cases, the impact on human workers is easy to see. What is more difficult to visualize is how similar technology could find its way into aspects of human work that are invisible and not as easily routable, such as complex decision-making, strategic planning and creative thinking.

Until recently, the consensus among researchers seemed to be that workers with a higher level of education would be less affected by automation than those lower in the economic hierarchy. Now, however, new research suggests that more educated white-collar workers may also face significant disruption. In a study published in November by the Brookings Institution, researchers found that certain higher-paying occupations may be more closely linked to A.I. technology in the future than expected. “The big point to remember is that professions and manufacturing professions will be strongly affected,” one of the study’s co-authors, Mark Muro, told me recently. “But so will white-collar workers – management and clerical activities.” Muro, senior researcher at Brookings’ Metropolitan Policy Program, who specializes in economic development and technology and wears glasses that give him a book look, noted that these jobs are likely to be replaced by AI or simply changed by that is not yet clear; in some cases, A.I. may end up helping human workers rather than doing their work for them. In any case, uncertainty is likely to cause the alarm in some circles.

One of the challenges of studying the effects of artificial intelligence on the employment of white-collar workers is that the integration of algorithms in office work is done slowly, and often imperceptibly. As a starting point, Muro and his co-authors tried to refine their task by focusing on machine learning, a form of AI. which uses algorithms to analyze huge amounts of data, find models, and then use those models to make predictions. “When most people talk about AI, they talk about it, in many ways,” said Muro. “When they talk about radiologists reading a scan improved by A.I., they are talking about machine learning.”


After defining what they were looking for, another challenge presented itself: how to determine which new types of machine learning applications were likely to be introduced in the coming years. For this, the researchers used a method developed by Michael Webb, a graduate student in the economics department of Stanford University, who created an algorithm to analyze AI the patents that have been filed and cross-check them with tasks performed. in various jobs. Webb examined a pool of about sixteen thousand patents containing verb-object pairs such as “diagnose disease” and “predict prognosis”, which corresponded to the descriptions of occupations used by the Ministry of Labor. “Patents are a reflection of what inventors think are important innovations that will make money in the future,” Webb told me. “The reason you patent something is that you think you could make money from this innovation, and you want to have the right to create this product and nobody does it instead.”

To test the effectiveness of this research method, Webb looked at around thirty years of patents in software and industrial robotics, to see if the predictions on employment and the drop in wages that we would have could find then had checked. They had: software patents often spoke of “recording”, “storage” and “information production”, while patents related to robots spoke of “cleaning”, “moving”, “welding” and of “assembly”. heavily on the tasks of packers and packagers, winch and winch operators, machine operators and those who worked in warehouses – for example, people driving forklifts. “It turns out that the jobs that were heavily exposed to these technologies have seen employment and wages decline over the next 30 years,” said Webb. This, for him, suggested that software and industrial robots were replacing human labor in these areas (although there are also other forces in force, such as the relocation of factories).

Webb then analyzed AI’s patent filings and found them using verbs such as “recognize”, “detect”, “control”, “determine” and “classify” and names like “models”, “images” and ” anomalies. ” The jobs that appear to deal with the intrusion of these new patents are different from the more manual jobs that were affected by industrial robots: intelligent machines can, for example, take on more tasks currently performed by doctors, such as detecting cancer, make predictions and interpret the results of retinal analyzes, as well as those of office workers who involve making data-based determinations, such as fraud detection or investigation of insurance claims. People with a bachelor’s degree may be more exposed to the effects of new technologies than other educational groups, just like those with higher incomes. The results suggest that nurses, doctors, managers, accountants, financial advisers, computer programmers and salespeople may see significant changes in their work. Occupations that require high levels of interpersonal skills appear to be the most isolated. (In particular, jobs at the top of the pay range, such as C.E.O., do not appear to have been significantly changed.)

When I asked Muro what he thought were the implications of his research, he told me about a tweet someone had written in response to the document’s findings. The author had expressed concern that his work would “muddy the waters for the underemployed and vulnerable,” said Muro, diverting the attention of truck drivers and factory workers who are already displaced from their work, often with less than a financial cushion. Fall back on. “There is no doubt that the white collar part of the story involves some of the most capable and resilient workers in the economy,” said Muro. “For someone in the eightieth percentile of income, they are probably well trained by their employers and invested by them.” This type of investment, added Muro, “does not go to those at the bottom of the line. ‘scale.’ But he noted that the danger of future automation for white-collar jobs could help make the issue more real for policy-makers. “I think it may broaden the scope of concern,” he said. “This suggests that it’s not just someone else’s problem, that is, a black or brown worker in a factory… We are all going to face huge flows and changes in the work and in our work. “