The strong instinct of most corporate leaders, faced with the teeming diversity not just of gender, race, and age but of thought, drive, and relationship inside their organizations, is to look for some way to exert control—to rein it all in, to impose conformity on the chaos, and thence to be able to understand what’s going on, and to shape what will happen next. And so companies have spent, and continue to spend, large quantities of time and money trying to work around each person’s uniqueness—and this is where these models bubble up from.

Humanyze, an MIT spin off led by Ben Waber, who coined the term people analytics, tags employees and leaders with sensors that capture their movements, communications, and even physiological responses (e.g., stress, excitement, and boredom). Just by analyzing anonymous group-level data, the firm can help organizations identify invisible elements of work relations, such as the hidden power dynamics, in a firm.

For example, in a recent study reported in Harvard Business Review, Waber and his team set out to decode the behavioral differences between men and women in a large multinational firm and explore whether such differences could partly explain the underrepresentation of women in the senior leadership ranks (where they accounted for just 20 percent). The researchers gathered email data, meeting schedule data, and location data for hundreds of employees, across all seniority levels, over four months. Of particular relevance was the data collected with sensors some employees wore. The sensors recorded who talked with whom; where, when, and for how long people communicated with each other; and who dominated each conversation. Waber’s team expected to find behavioral differences between men and women pertaining to people’s drive and networking habits: ‘Perhaps women had fewer mentors, less face time with managers, or weren’t as proactive as men in talking to senior leadership.’ However, the results showed no significant differences between what women and men did at work: ‘Women had the same number of contacts as men, they spent as much time with senior leadership, and they allocated their time similarly to men in the same role. We couldn’t see the types of projects they were working on, but we found that men and women had indistinguishable work patterns in the amount of time they spent online, in concentrated work, and in face-to-face conversation. And in performance evaluations men and women received statistically identical scores. This held true for women at each level of seniority. Yet women weren’t advancing and men were.

Their [Frank Dobbin and Jiwook Jung] results showed that although adding more women to boards did not change the firms’ performance, it led to a decrease in the firms’ stock valuation. These findings highlight an alarming reality: irrespective of the actual performance differences between men and women, people—and, in this case, investors—are unlikely to change their beliefs, and beliefs drive decisions.

Over the following year, Donald Trump’s spectacular rise and win in the 2016 US presidential election would shock the left-leaning workforce of the tech industry into self-reflection. As upheaval ripped through companies like Meta and Google and techlash sentiment gripped the public, AI researchers, too, began to question whether the field had moved too quickly to yoke its technologies to corporate bottom lines. An accounting of the societal impacts of commercializing AI research returned an unsettling scorecard: Automated software being sold to the police, mortgage brokers, and credit lenders were entrenching racial, gender, and class discrimination. Algorithms running Facebook’s News Feed and YouTube’s recommendation systems had likely polarized the public, fueled misinformation and extremism, enabled election interference, and, most horrifying in the case of Facebook, precipitated ethnic cleansing in Myanmar.

At a time when freewheeling corporate research was still permitted, Microsoft anthropologist Mary L. Gray and computational social scientist Siddharth Suri were among the first to show the world the plight of workers like those in Kenya who build their livelihoods around an essential piece of the AI supply chain.

In 2019, they published their book Ghost Work, based on five years of extensive fieldwork, revealing a hidden web of piecemeal labor and digital exploitation that propped up Silicon Valley. Tech giants and unicorns were building their extravagant valuations not just with engineers paid six-figure salaries in trendy offices. Essential, too, were workers, often in the Global South, being paid pennies to carefully annotate reams of data. Take self-driving cars. A self-driving car needs to drive in the correct lane, respond to erratic driving behavior, and pause at a safe distance for schoolchildren crossing the road. To do this, the software system controlling the car uses an amalgamation of several deep learning models, including those dedicated to recognizing objects on the road: lane markings, road signs, traffic lights, vehicles, trees, pedestrians.