While theological perspectives remain open to the possibility of artificial intelligence (AI) developing the capacity for religious experiences, one theologian argues that the deeply embodied and evolutionary roots of human religion suggest that even highly advanced AI systems would likely be incapable of developing anything comparable to human religious behavior. Marius Dorobantu, an Assistant Professor of Theology and Artificial Intelligence at the Free University of Amsterdam, makes this argument in a recent article in Zygon: Journal of Religion and Science (online Oct. 25). He starts by discussing the case of a Google engineer who claimed in 2022 that an AI chatbot he was testing had become sentient. His belief was partly fueled by the program’s mention of God, its expressed fear of being turned off, and its declaration of being a “spiritual person.” Dorobantu warns that it is not difficult today for chatbots to generate statements outlining their interest in religion if prompted to do so. But this does not make them authentically religious—and authenticity is essential here. Notwithstanding “our inherent tendency to anthropomorphize AI,” and assumptions in science fiction that intelligent robots “will develop very humanlike concerns, fears, and desires,” even “the hypothetically superintelligent AIs of the future, endowed with artificial general intelligence, will likely be profoundly different kinds of entities from us.”
Source: @dinolover8558 (AI Emojis)-
Dorobantu notes that even if AI systems achieve human-level intelligence, their internal structure, world-modeling, and problem-solving approaches remain radically different from our own. This fundamental difference casts doubt on whether AI would develop the same kind of existential anxieties and spiritual needs that drive humans to religion. He questions the widespread bias toward equating religion primarily with its intellectual or doctrinal aspects. People assume robots would develop religious tendencies mainly through intellectual curiosity about existence and causality. This overlooks the deeply embodied and experiential nature of human religion. “If AI programs start to declare an adherence to certain religious beliefs, this will equate to only the upper, and arguably more superficial, layer of how religion functions in humans, thus lacking the emotional embedding and psychological structure that would grant such claims plausible authenticity.” Still, Dorobantu admits that his conclusion is merely tentative—“it only refers to a type of religiosity very similar to our own, which is the only way we can so far imagine.” He maintains openness to the possibility of different forms of religious experience emerging through artificial intelligence, while emphasizing the need for careful evaluation of such claims based on improved understanding of both human religion and artificial intelligence.
(Zygon, https://www.zygonjournal.org/)
