We are also doing AI Phrenology

Every few months, a brilliant student asks: What if I tried to do this kind of 18th century science, but with AI?

The the latest entry entering AI phrenology history comes from a group of economists who claim to have developed a method to analyze a single image of a person’s face in order to calculate their personality and predict what they will do in education and work.

Some recent studies in AI phrenology– such as algorithms that claim to predict a person’s sexuality or likelihood of committing a crime based on facial expressions – a lot to be criticized and debunked. Research has also shown that commercial AI tools that claim to measure personality traits are not very reliable.

However, Marius Guenzel and Shimon Kogan, of the Wharton School of the University of Pennsylvania; Marina Niessner, of Indiana University; and Kelly Shue, from Yale University thought that the image of a person’s face can determine their personality. They received funding for their research from several AI and financial research funds at Wharton and have presented their findings at financial conferences and universities around the world, according to their paper.

The authors took LinkedIn photos of 96,000 MBA graduates and manipulated facial expressions that they say measure how well a person would score on the Big Five personality test, which rates people for their openness, conscientiousness, extroversion, agreeableness, and. neuroticism.

They then measured the correlation between the number of removed personalities and the prestige of the MBA program they completed and their salary (as estimated by employers analyzing LinkedIn data).

Based on this analysis, the authors concluded that personality plays a “significant role” in predicting whether a person will attend a school with a top MBA program and how much they will earn in their first job after graduation. For example, Men who are in the top 20 percent of the “important” personality went to MBA programs had 7.3 percent higher and had 8.4 percent more money than men whose personality was below the 20 percent of the important. When the researchers controlled for factors such as a person’s race, age, and attractiveness (all taken into account), the results were limited.

In particular, the authors seem to have not tested themselves to ensure that the Big Five personality traits their algorithm derived from LinkedIn headshots were correct. None of the people whose photos were analyzed attempted the Big Five personality test to confirm the algorithm’s results.

The professors wrote that their findings show “the important role of non-cognitive skills in shaping work outcomes” and that the use of AI to analyze faces, instead of giving human tests to people, “provides new avenues for academic inquiry … (and calls for) a re-examination of ethical principles.” good, useful, and intelligent that support the use of such technologies.”

At the same time, they wrote that the method they have just shown should not be used to analyze the labor market and that “depersonalization represents statistical discrimination.”

In other words, the scientists stopped to think about whether they should do it, said it was discrimination, and then did it anyway.