Appiah Warns AI May Erode Moral Character and Autonomy

Philosopher Kwame Anthony Appiah, in a lengthy essay for Humanist Review, argues that AI's impact on moral character demands attention beyond mere performance metrics. Drawing on John Stuart Mill's On Liberty, Appiah warns that technologies extending human powers can erode the faculties they assist. He cites studies showing AI improves colonoscopy adenoma detection but may deskill physicians. Appiah emphasizes Mill's distinction between what people do and what manner of persons they become, arguing that autonomy, individuality, and active cultivation of faculties are constituents of a good life. He critiques moral deference to AI, noting that even if LLMs give excellent moral answers, users should understand reasons behind them. Appiah discusses hidden assumptions in AI, including political biases from training data and fine-tuning, and warns against sycophantic AI that affirms users excessively. He cautions against synthetic intimacy replacing real relationships, citing Iris Murdoch's insight that love requires recognizing another's reality. Appiah proposes that AI could enhance humanity if designed to encourage deliberation rather than deference, but constitutive de-skilling and erosion of character remain risks. He concludes by invoking Mill's own recovery from breakdown through reading Wordsworth and his partnership with Harriet Taylor, urging that the question is not just what AI enables but what manner of people we become.