Sexual Autonomy in the Age of Artificial Intelligence

• In Event: Searching for the Human: Technology, Democracy, and the Politics of Modernity

Abstract

Artificial intelligence is coming, and with it numerous questions about its meaning for the future of humanity. While existential risk predominates our conversation—what if AI results in the end of humans altogether?—no less important are the smaller, more prosaic and yet deeply profound challenges to our human nature which accompany every technological shift. In the age of artificial intelligence, like the age of the internet, and before that the printing press, we must discover how best to reconcile our new technological capacities with our old human frailties. And nowhere is this more true than in the realm of sex: that most beastly, most angelic, most human of all activities.

The importance of sex to the human condition has led to the creation of numerous social and legal norms meant to protect the value of this area of life, among which “sexual autonomy” is an increasingly important governing concept today. We rely on sexual autonomy in carving out privacy from the eyes of the state, in pursuit of equality in gender relations, and in the creation of norms that enable physical and emotional integrity. Yet sexual autonomy is thus far a neglected concept in the discourse around artificial intelligence, leaving open a dangerous lacuna: how will AI impact human sexuality?

This paper examines one particular threat to sexual autonomy by artificial intelligence: that of algorithmic artificial pornography. AI-generated sexual media may threaten human sexual autonomy in two ways. First, we may have reason to be worried about sexual manipulation of consumers by companies via their bots. If the AI algorithm is attempting to predict consumer preferences, it may guide the consumer towards forms of sexual media that he or she would not otherwise have chosen, disempowering the consumer as a sexual agent. While some consumers may experience such algorithmic guidance as beneficial, exposing them to new and perhaps ultimately more desirable forms of sexuality, we may also have reasons to worry about potential harms. If sexuality is at all habituated, meaning that our pornography impacts our desires—a matter of some controversy—then having machines “choose” our sexual experiences may undermine our capacity for human sexual flourishing.

Second, and still more worrisome, AI promises the potential of non-human “interactive” online sexuality, such as AI partners or scenes in “virtual reality” (VR). Interactive sexual media has the potential to bring many benefits to consumers, especially for those who have struggled to find partnerships in the physical world. However, here we have the problem that the consumer is not the only active participant in determining what sexual acts are pursued; because the AI can act in response to the consumer, the bot or VR-partner can behave in ways which compromise the sexual autonomy of the human. What should our response be to rapes in virtual space? We have some research on this behavior, but only with regards to other human actors (Horne 2023); what moral agency can we ascribe to the AI, and what might we owe victims if the system does behave in ways experienced as a violation?

Overall, this paper argues that we ought to institutionalize protections for sexual autonomy into our systems for governing artificial intelligence, and posits three means of doing so. First, via the democratic process of legislation, we ought to mandate that producers of artificial intelligence meet certain standards of interpretability and explainability to enable consumer control over their sexual media. Second, via the infrastructures of the technologies themselves, we must provide safety mechanisms that enable companies to understand when and how AI tools act in ways that harm sexual autonomy. Third and finally, via corporate governing structures, we must determine a bill of sexual rights to which consumers are entitled, ensuring sexual autonomy not only for current AI tools, but for those we develop in the future.

Author

• Joan Eleanor O'Bryan, Stanford University