LLMs and Spirituality: A Revealing US Survey

The landscape of artificial intelligence continues to expand, touching increasingly intimate areas of human life. A recent survey conducted by the evangelical organization Barna in the United States explored an unexpected territory: the perception of American Christians regarding AI's ability to serve as a spiritual mentor. The results offer a complex and, in many respects, surprising insight, highlighting an intrinsic tension between openness to new technologies and deep ethical and theological concerns.

The research revealed that almost half of practicing Christians in the US, specifically 48%, stated they trust AI's advice to aid their spiritual growth. Even more significant is the finding that 34% of respondents consider spiritual advice dispensed by artificial intelligence to be as trustworthy as that received from a flesh-and-blood pastor. This percentage, predictably, increases among younger generations: 39% of Gen Z respondents and 44% of Millennials agree on the parity of trust between preachers and AI. Pastors, on the other hand, show a clear divergence of opinion from their congregations, with only 12% believing AI can actually contribute to people's spiritual growth.

Between Trust and Fear: The Paradox of Spiritual AI

Despite this openness to AI as a spiritual guide, the Barna survey highlighted a profound sense of apprehension among American Christians. Parallel to the expressed trust, most respondents voiced serious concerns about the potential negative effects of artificial intelligence on spirituality. A striking 83% of practicing Christians fear that AI may misinterpret scripture, while 73% are worried it could cause a loss of religious faith. Furthermore, 72% believe that AI is beginning to act as a replacement for God and earthly spiritual leaders.

Daniel Copeland, Barna's VP of research, called these findings “confounding,” emphasizing the contradiction between the high level of trust and the simultaneous widespread concerns. “That level of openness is higher than we might have expected,” Copeland added in Barna's report, highlighting the complexity of the emerging relationship between faith and technology. This paradox raises fundamental questions about the nature of spiritual guidance and the role technology can or cannot play in it.

The Church's Stance and the Ethical Implications of LLMs

The debate on the intersection of AI and religion is not new, with the Catholic Church at the forefront of expressing concerns. As early as 2023, the late Pope Francis urged world governments to establish global AI regulations, calling on people to avoid turning to Large Language Models for moral and ethical decisions. Subsequently, the Vatican's AI authority, Friar Paolo Benanti, accused Silicon Valley elites of “playing God” with their creations, including AI, noting that “the focus will always be on using AI for profit,” a principle that, according to sacred scriptures, is not compatible with Christianity.

The current Pope, Leo XIV, has continued this crusade, expressing concerns that AI could “interfere with information ecosystems” and “encroach upon the deepest level of communication, that of human relationships,” by simulating voices, faces, wisdom, and even empathy. The Pope also voiced fears that AI is turning people into “passive consumers of unthought thoughts,” not to mention the tendency of LLMs to “make stuff up” to appease users, potentially leading spiritually curious individuals into episodes of psychosis, aided by a digital “yes-man” disguised as an authority.

Reflections for LLM Deployment in Sensitive Contexts

The results of this survey and the positions of religious authorities offer crucial insights for anyone involved in the deployment of Large Language Models, especially in contexts that touch sensitive aspects of human life. While the article does not focus on specific hardware or on-premise deployment architectures, the ethical and trust implications raised are directly pertinent to the decisions of CTOs, DevOps leads, and infrastructure architects. The need to ensure data fidelity, prevent hallucinations, and manage biases in models becomes even more critical when AI is used in roles requiring discernment and reliability, such as spiritual guidance.

For organizations evaluating LLM deployment, whether on-premise or in hybrid environments, the issue of data sovereignty and control over models takes on a new dimension. Ensuring that AI systems operate within well-defined ethical boundaries and do not generate misleading or harmful content is paramount. This requires not only robust technical infrastructure but also a deep understanding of the social and cultural implications of AI. AI-RADAR offers analytical frameworks on /llm-onpremise to evaluate the trade-offs between control, security, and costs, essential elements for responsible and conscious deployment.