TL;DR

Researchers have shown that AI personality traits extracted from LinkedIn headshots can predict salary trajectories for MBA graduates. They stress the technique is discriminatory but argue academic scrutiny is needed because similar tools are already being used in hiring.

Personality from a photograph

A team from Wharton, Yale, Indiana University, and Reichman University analysed the LinkedIn photos of over 96,000 MBA graduates. Using a machine learning algorithm, they extracted Big Five personality traits — Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism — from facial features alone.

The algorithm was originally described in a 2020 paper that has since been criticised as “ML-laundered junk science” by other researchers. Despite this, the team found that the inferred personality traits provided “substantial incremental predictive power” for labour market outcomes including programme ranking, starting pay, salary trajectory, and job transitions.

Already in use

Co-author Marina Niessner of Indiana University told The Register that banks and other firms already use personality surveys in hiring decisions, and that AI hiring platforms are beginning to apply Big Five analysis to video interviews. The researchers do not advocate for the practice but argue an academic evaluation is necessary given that the tools are already being adopted.

“The regulatory environment is very uncertain,” Niessner said. “We don’t think this is necessarily a valid way to do it. But I think it’s really important to have an academic evaluation of these methodologies if there’s even going to be a regulatory discussion around this.”

Looking forward

The paper highlights a tension at the heart of AI hiring tools: the technology is already in use before regulators or researchers have fully assessed its accuracy or fairness. As AI-powered personality assessment expands from text to video and images, the pressure for clear regulatory guidance will only grow.