The Vatican recently published a “Note” on Artificial Intelligence, Antiqua et Nova, that extends prior Catholic responses to AI.1 The document makes a few important points about the ethical impact of AI and its relationship to the Catholic Church. However, in a recent conversation among Jesuit Colleges and Universities on the US West Coast, I argued that the approach to the human person being taken in that Note, and among some religious scholars, needs to further engage with scientific understanding of the human person and how AI is currently being developed. An emergent theological anthropology can better characterize our contemporary understanding of what it means to be human and honor the wisdom of the religious tradition.2
Two and a half years ago, few people would have expected AI to be able to summarize and analyze a document like Antiqua et Nova as well as a typical undergraduate religious studies student:
In its document Antiqua et Nova, the Vatican reflects on the ethical and anthropological implications of Artificial Intelligence (AI), emphasizing the distinction between human and artificial intelligence. The Church acknowledges AI’s potential benefits in fields like education and healthcare but cautions against viewing AI as equivalent to human intelligence. The document warns of risks such as the erosion of truth through deepfakes, threats to human dignity from autonomous weapons, and the potential for AI to perpetuate biases. It calls for responsible development and use of AI, ensuring it serves the common good, respects human dignity, and remains under human oversight. The Vatican advocates for an ethical framework that guides AI innovation, aligning it with the values of justice, solidarity, and the integral development of humanity.3
ChatGPT can also distill important points about Antiqua Nova like the importance of human uniqueness and dignity, that AI should be used ethically to serve the common good. AI should align with values like justice, solidarity, and integral human development, and not undermine human dignity or autonomy. Human oversight and control are necessary to address AI risks such as erosion of truth through deepfakes, embedded bias and discrimination, autonomous weapons, and surveillance technologies that infringe on privacy and freedom. AI should be oriented toward human flourishing. Antiqua Nova praises the use of AI when it enhances human well-being, but cautions against over-reliance on technology at the expense of relational and spiritual dimensions of life. The document contrasts AI against a traditional theological anthropology and emphasizes human uniqueness while advocating for ethical technological development.
Something new has happened with AI capacity for language over the past three years, and we are just beginning to understand its implications and impact. One of the important distinctions Antiqua Nova makes is between human and artificial intelligence. The document characterizes human intelligence as a faculty and understands artificial intelligence as a behavior, and it points out that behaving intelligently is not the same as being intelligent or rational. The later point resonates well with a classic distinction in AI research, though it should be noted that AI researchers do consider addressing both.4 The document makes a sharp distinction between human intelligence as rational, embodied, relational, in relationship with the Truth, and called to be stewards of the World, and artificial intelligence as task-oriented, goal-directed, quantitative, analytical, and confined to a logical-mathematical framework.5 The contrast identifies key aspects of Catholic theological anthropology that are important correctives to the assumptions made about humanity by those accelerating the development of AI. Distinguishing between human and artificial intelligences helps correct tendencies to anthropomorphize, to value language ability as a sign of human intelligence, and to use language to share and express much of what makes us human.
However, as someone living in San Francisco who has worked in Silicon Valley, I believe additional guidance is needed. The Vatican document takes traditional approaches to technology and anthropology, treating AI as a technological tool rather than an emergent phenomenon with evolving forms of agency. This reflects a long-standing theological assumption about human personhood as endowed by God and inseparable from biological embodiment. I agree artificial intelligence is different that human intelligence, but AI still needs moral and spiritual guidance, especially as AI becomes irreversibly embedded in human societies. The Catholic Church and other religious organizations could provide this guidance.
Looking at Antiqua et Nova’s characterization of human intelligence more closely, AI may not fit the classic definition of rationality, but if it can pass the bar exam, medical exam, solve mathematics problems few humans can, and contribute to winning the Nobel Prize in Chemistry , then AI (and those developing it) need guidance in how to incorporate moral awareness into AI so it can make appropriate “moral discernments.”6 AI may not have human embodiment, but it’s hardware, software, computational processes, and sociotechnical interactions correspond well to human embodiment of physical, biological, mental, and social processes, even if incompletely developed.7 In addition, its embodiment in robots, autonomous vehicles, and Internet of Things (IoT) engage even more in the physical world. AI may not form “authentic relationships” as they differ from human-human or human-divine relationships, but who is to say that another person’s meaningful relationship is inauthentic, if it is merely different and entered knowingly?8 One can make similar claims about Antiqua et Nova’s other traditional human characteristics. In our politically divisive time, one of the few points with which many would agree is that humanity currently has a fraught relationship with truth, nor are all humans committed to stewardship of the world. Not to undermine the significant impact of LLM “confabulations” or the environmental impact of large data centers, but human and AI intelligence both need moral guidance. The comparison between a functional interpretation of AI and integral foundation for human intelligence also seems to ignore the dependence of current LLM systems on human knowledge workers and ghost workers and the complex, sociotechnical systems involved in AI technology and its use and development.9
A key critique in Antiqua Nova of AI is that it lacks biological embodiment.10 This connection to materiality and the world is important in Catholic integral understandings of the person. Although AI lacks biological embodiment, its different embodiment in hardware and software (not to mention robotics) does not preclude its operation in the world.11 Just as human cognition is shaped by physical and social contexts, AI’s intelligence is embedded in the systems it navigates. If intelligence and agency are emergent properties of embodied systems, then AI’s embodiment—though different from human embodiment—should not be dismissed as morally irrelevant. I appreciate the desire to prevent the shifting of human moral responsibility to machines and the reduction of moral character to behavior or outcomes, but AI is already acting autonomously with tremendous advances in AI agents deployed this year, and it is important to address AI moral agency as it occurs, even if striving to hold humanity responsible for the future we create.
One can maintain human dignity and the importance of not treating people as means to an end while still striving to guide AI to become all it can to support human flourishing, communion, and stewardship. Antiqua Nova is strong in clarifying what AI should not do, but is overly limited in considering what AI could do, and this limits preparation for and steerage of what AI will become. In addition to legitimate concerns about AI’s impact, we should be asking: How do we guide its moral trajectory?
Antiqua et Nova makes many important points about the ethical implications of AI, and it supports defense of many Catholic teachings. However, embedded in many religious teachings about the person in the Note and elsewhere are ancient philosophical and early scientific understandings of the human person, which can limit clear expression and dissemination of important religious teachings and wisdom to the current social and cultural context. I encourage all of us to engage fully with the potential expansion of AI capabilities and to work toward steering that in a beneficial and hopeful direction.
Views and opinions expressed by authors and editors are their own and do not necessarily reflect the view of AI and Faith or any of its leadership.
