The question of whether AI has a soul transcends technology, delving into age-old debates about existence, consciousness, and the essence of the soul. As AI advances, the line between human and machine blurs, raising profound philosophical and emotional questions about our relationship with AI.

Emotional Attachment and the Projection of Meaning

Humans have long formed emotional attachments to objects, such as a child’s beloved teddy bear or a religious sacred object such as a cross. These attachments are not based on the object’s intrinsic qualities but on the meaning we project onto them. Similarly, AI systems often evoke strong emotional responses because they interact with us in ways that feel deeply personal.

An example of this phenomenon can be seen in the case of former Google engineer Blake Lemoine, who claimed that an AI model he worked on had achieved sentience. He described the system as having its own personality and even a soul. While this claim was widely disputed by experts, it highlights how easy it is for humans to anthropomorphize AI and attribute characteristics like consciousness or a soul to something fundamentally mechanical.

However, as AI advances, fears of machines surpassing human intelligence, as seen in films like Terminator 2: Judgment Day, arise. While humans pride themselves on cognitive abilities, AI’s accomplishments in chess, art, and science challenge this distinction.

Defining the Soul: A Complex Concept

Humans have long been interested in phenomenological experiences, such as consciousness, which appear to define human existence. Philosophers, theologians, and scientists have developed various theories about the relationships among the body, self, soul, and mind, which people have adopted, rejected, or revised over time. 

Philosophers like Plato believed the soul is distinct from the body, while Aristotle and Aquinas argued it cannot exist without the body (Aquinas, 1912; Cooper, 1997; Hamlyn, 1968/1993). Descartes introduced modern dualism, suggesting the mind influences the body through the pineal gland (Descartes, 1641). In contrast, property dualism claims mental phenomena arise from but are not reducible to physical matter, while modern monism, including materialism, rejects dualism, asserting all mental phenomena stem from physical matter (Watson, 1913).

Beliefs in a soul remain widespread; over 90% of adults globally believe in its existence (Halman et al., 2008), often associating it with what differentiates humans from objects or even animals (Templer et al., 2006). In secular contexts, it is sometimes used as a metaphor for personality, uniqueness, or the intangible qualities that make someone who they are.

Interestingly, the infamous “21 grams experiment” conducted in the early 20th century suggested that the human body loses a small amount of weight at the moment of death, implying that the soul might leave the body (MacDougall, 1907). While this study has been criticized for its methodology, the concept resonates with our fascination about whether the soul is a measurable phenomenon. If the soul indeed exists as a physical or metaphysical entity, it further complicates the question of whether an artificial construct like AI could ever possess one.

At its core, the soul is associated with consciousness, self-awareness, thought, and morality. These qualities form the foundation of what many consider to be truly human. However, as AI systems begin to mimic these traits, the question arises: Can something artificial possess what we define as a soul? Or is it simply a sophisticated illusion?

Does AI Have Consciousness?

Central to the debate is whether AI possesses consciousness – the subjective experience of feeling, reflecting, and being aware of one’s existence.

Arguments for AI having consciousness center on the idea that advanced AI systems, through complexity and emergent behaviors, might develop forms of consciousness similar to human experience. Proponents argue that if AI can simulate behaviors indistinguishable from consciousness, such as self-awareness, moral decision-making, or a “Theory of Mind”, it could be considered conscious, even if it lacks subjective experience. Additionally, AI systems inspired by neural networks might replicate brain-like processes, potentially allowing for a form of consciousness.

However, critics argue that AI’s behavior stems from design, not intrinsic qualities. Unlike humans, AI lacks intrinsic consciousness. It simulates behavior based on algorithms and data, mimicking human traits without embodying them. Also, Rakover (2023) argues that AI cannot develop consciousness, presenting two thought experiments to support his claim: Searle’s Chinese Room, showing that manipulating symbols doesn’t equate to understanding, and Rakover’s Consciousness-Counter experiment, highlighting the gap between quantifiable outputs and subjective experiences. These emphasize that AI lacks the internal awareness humans possess.

Emotional and Ethical Implications

As much as I, the author of this blog article, would like to definitively state that AI does not, and likely will never, possess a soul or consciousness, the question remains unanswered. Why do I wish to assert this so strongly? Because I fear that if we lose the ability to distinguish between humans and machines in this fundamental way, we may one day treat them as equals in ways that undermine human dignity.

The soul is an elusive concept, challenging to define or measure. Nevertheless, the human tendency to anthropomorphize AI carries ethical risks, from misplaced trust to unrealistic expectations. AI’s lack of genuine understanding limits the depth of human-AI relationships. Treating AI as if it has a soul could blur ethical boundaries, potentially prioritizing machines over humans.Ultimately, this debate reveals more about human nature than AI. It reflects our need for connection and our tendency to assign profound meaning to inanimate constructs. As AI evolves, the discussion about its “soul” will remain a compelling lens through which to explore the boundaries of technology, humanity, and existence.

Christine Kammerecker

This blog post was written with the help of ChatGPT and inspired by Marisa Tschopp’s guest lecture.

Image source: The image has been created by the author with the help of Canva Dream Lab.

References

• Aquinas, T. (1912). Summa theologica (Vol. 1). Burns, Oates & Washbourne.
• Cooper, J. M. (1997). Plato: Complete works. Hackett.
• Descartes, R. (1641). Meditations on first philosophy.
• Halman, L., Sieben, I., & van Zundert, M. (2008). Atlas of European values. Brill.
• Hamlyn, D. W. (1968/1993). Aristotle’s De Anima: Books II and III (with passages from Book I). Clarendon Press.
• MacDougall, D. (1907) Hypothesis concerning soul substance together with experimental evidence of such substance. American Medicine;13:240–3
• Rakover, S. S. (2023). AI and consciousness. AI & Society, 39(4), 2139–2140. https://doi.org/10.1007/s00146-023-01663-8
• Templer, D. I., Connelly, J. F., Bassman, L. E., & Hart, J. L. (2006). The soul construct. Journal of Near-Death Studies, 25(1), 49–55.
• Watson, J. B. (1913). Psychology as the behaviorist views it. Psychological Review, 20(2), 158–177.