[...] Their research is another expression of Silicon Valley’s fake-it-til-you-make-it culture of denial and opportunism. Yet the outsize hype attached to these projects breeds problems of its own. Regardless of machine-learning “gaydar’s” efficacy, James Vincent wrote at The Verge, “if people believe AI can be used to determine sexual preference, they will use it.” It follows that we’re better served by drawing attention to the error rates and uncertainties of this research than by obscuring its flaws and echoing the bluster of its pitchmen. In the end, Kosinski got it wrong: if he wanted to help the queer community fight “oppressive regimes,” he might have led the publicity drive by trumpeting his paper’s shortcomings. We know from past hokum, like the Myers-Briggs Type Indicator, handwriting analysis, and the polygraph, that the public and private sectors will bet on coin-toss odds. This technology is attractive, despite its failures, because it offers an illusion of standardization and objectivity about that which is conditional and even subjective.

If technology is routinely legitimized by delusions about its impartiality and misplaced faith in its precision, perhaps a wider public acknowledgment of its capacity to fail might slow its unrelenting advance. The failures of surveillance and classification technologies, frustrating as they might be in the moment—especially for their investors—cast doubt on the powerful, the knowledgeable, and the expert. These mistakes demonstrate that systems do not work as intended. And these founderings might also give way to conversations that take place beyond the noisy gibberish of marketing language. Considering its false advertisement, how is this technology used? What restrictions should be placed on it? Should this technology exist? At the risk of creating my own blunt “Strong Female Lead”-style categorization: I think a technology should not exist if there is no procedure to contest and amend its inevitable mistakes. And that’s just to start.