Every few months, another highly educated academic asks: What if I tried doing debunked 18th-century race science, but with AI?
The latest entry into the AI phrenology portfolio comes from a group of economics professors who say they’ve developed a method for algorithmically analyzing a single photo of a person’s face in order to calculate their personality and predict their educational and career outcomes.
Other recent academic forays into AI phrenology—like algorithms that purport to predict a person’s sexuality or the likelihood they will commit a crime based on their facial features—have been widely criticized and debunked. Investigations have also shown that commercial AI tools that claim to measure personality traits are extremely unreliable.
Nonetheless, Marius Guenzel and Shimon Kogan, of the University of Pennsylvania’s Wharton School; Marina Niessner, of Indiana University; and Kelly Shue, from Yale University decided that a snapshot of a person’s face can determine their personality. They received funding for their research from several AI and finance research funds at Wharton and have presented their findings at financial technology conferences and universities all over the world, according to their paper.
The authors collected the LinkedIn profile pictures of 96,000 MBA program graduates and ran them through a facial analysis algorithm that allegedly measured how the person would score on the Big Five personality test, which rates people on their perceived openness, conscientiousness, extraversion, agreeableness, and neuroticism.
They then measured the correlation between these extracted personality scores and the prestigiousness of the MBA program they completed and their eventual compensation in the workforce (as estimated by a proprietary model that analyzes LinkedIn data).
Based on this analysis, the authors concluded that personality plays an “important role” in predicting whether a person will attend a school with a highly ranked MBA program and how much they will earn in their first job after graduation. For example, Men in the top 20 percent of “desirable” personalities attended MBA programs ranked 7.3 percent higher and had estimated incomes 8.4 percent higher than men whose personalities were in the bottom 20 percent of desirability. When the researchers controlled for factors like a person’s race, age, and attractiveness (all of which were inferred), the effects became smaller.
Notably, the authors don’t appear to have made any independent effort to establish that the Big Five personality scores their algorithm extracted from LinkedIn headshots were accurate. None of the people whose profile pictures were analyzed took a Big Five personality test to confirm the algorithm’s conclusions.
The professors wrote that their findings highlight “the critical role of non-cognitive skills in shaping career outcomes” and that using AI to analyze faces, rather than actually administering personality tests to people, “presents new avenues for academic inquiry … [and invites] further exploration into the ethical, practical, and strategic considerations inherent in leveraging such technologies.”
At the same time, they wrote that the technique they just demonstrated shouldn’t be used for labor market screening and that “personality extraction from faces represents statistical discrimination in its most fundamental form.”
In other words, the scientists did stop to think about whether they should, concluded it was discriminatory, and then did it anyway.