Georgetown Law
Technology Review

Suggesting that it is possible to determine a person’s interior characteristics or future social outcomes based on their facial expressions, body movements, and other characteristics is not backed by scientific consensus.1See, e.g., Lisa Feldman Barrett, Ralph Adolphs, Stacy Marsella, Aleix M. Martinez, & Seth D. Pollak. Emotional Expressions Reconsidered: Challenges to Inferring Emotion From Human Facial Movements, 20 Psychol. Sci. Pub. Int. 1, 68 (2019) (concluding in meta-study that it isn’t possible to judge emotion by just looking at a person’s face); see also Angela Chen & Karen Hao, Emotion AI Researchers Say Overblown Claims Give Their Work a Bad Name, MIT Tech. Rev. (Feb. 14, 2020), https://www.technologyreview.com/2020/02/14/844765/ai-emotion-recognition-affective-computing-hirevue-regulation-ethics [https://perma.cc/L5FR-FJ3Y]. Technologies that aim to do so often reflect discredited and racist pseudoscientific practices, including physiognomy, phrenology, and other forms of race science.2See, generally Luke Stark & Jevan Huston, Physiognomic Artificial Intelligence, Social Science Research Network, (Sep. 24, 2022 ), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3927300; Luke Stark, Facial Recognition Is the Plutonium of AI, 25 XRDS 50, 52, 55 (2019) https://dl.acm.org/doi/pdf/10.1145/3313129 [https://perma.cc/D2RZ-K4YC]; see, e.g., Blaise Aguera y Arcas, Margaret Mitchell, & Alexander Todorov, Physiognomy’s New Clothes, Medium (May 7, 2017), https://medium.com/@blaisea/physiognomys-new-clothes-f2d4b59fdd6a (describing usage of outer appearance to infer internal character) [https://perma.cc/64G4-GA9V?type=image]; Sahil Chinoy, The Racist History Behind Facial Recognition, N.Y. Times (Jul. 10, 2019), https://www.nytimes.com/2019/07/10/opinion/facial-recognition-race.html (describing physiognomy and phrenology respectively as using “facial structure and head shape to assess character and mental capacity.”) [https://perma.cc/K5U2-KSLQ]; see also Sam Biddle, Troubling Study Says Artificial Intelligence Can Predict Who Will Be Criminals Based On Facial Features, Intercept (Nov. 8, 2016) (“phrenology [is] just using modern tools of supervised machine learning instead of calipers. It’s dangerous pseudoscience’”). These practices interpret physical differences between people as signs of their inner worth and character, and use this to justify oppression, subjugation, and inequality.3See Stark, supra note 11, at 53 (“In the case of facial recognition, the schematization of human facial features is driven by a conceptual logic that …theorists have identified as fundamentally racist because it is concerned with using statistical methods to arbitrarily divide human populations.”). Critical race scholars continue to articulate the connections between systems of racial oppression and quantification. See, e.g., Kimberlé Crenshaw, Mapping the Margins: Intersectionality, Identity Politics, and Violence Against Women of Color, 43 Stan. L. R. 1241, 1299 (1991).

Continue reading