Experiments in Artificial Intelligence

In artificial intelligence (AI) research and development, a benchmark combines data and evaluation criteria to measure how well AI systems perform specific tasks or demonstrate particular abilities. Developing meaningful benchmarks represents a significant opportunity for religious and ethical scholars to shape AI’s future in a positive direction. With Generative AI, evaluation methods alone can orient AI training processes toward moral improvement and religious awareness. Recent efforts to evaluate the ethical dimensions of large language models (LLMs) set the stage for extending this frontier as do well-recognized challenges in constructing models that capture diverse human cultures. Developing benchmarks that more broadly measure human and other suffering, compassionate responses, and richer conceptions of flourishing and well-being provide timely and impactful means to affect AI development. Efforts toward these state-of-the-art ethical and religious benchmarks are reviewed, discussed, and situated within a broader framework oriented toward flourishing.

This paper proposes an open, ontology-driven, AI-mediated research environment for the study of Greek religious texts built from public-domain and openly licensed corpora rather than proprietary platforms. Using TEI/XML resources such as GLAUx, Open Greek & Latin, and Perseus/Scaife, the project explores whether agentic coding tools can automate corpus normalization, metadata reconciliation, citation alignment, hyperlink generation, ontology construction, and graph-based retrieval across biblical, early Christian, and related Greek materials. Unlike conventional search systems centered on keywords, lemmas, or fixed indices, the proposed environment supports relation-aware discovery, customizable semantic linking, and provenance-preserving exploration across corpora. Rather than treating AI as an autonomous interpreter, the paper argues that agentic systems function as infrastructural and hermeneutical assistants: they lower the cost of building open research environments and help scholars navigate, explain, and extend them while preserving human judgment as decisive.

This paper introduces the InterSapience Project, a theological and interreligious proposal for rethinking AI alignment as a moral and relational problem rather than merely a technical one. I advance the notion of “co‑intelligence” (the integrated dynamics of collective, collaborative, and cooperative intelligence) as a normative framework for human–synthetic cooperation. Drawing on process thought, comparative theology, and interfaith ethical resources, the study reframes alignment as relational resonance and mutual transformation. It argues that the crucible of AGI/SSI demands practices of humility, dialogical encounter, and structural justice that theological traditions already model. By engaging AI scholarship (Bostrom, Russell, Tegmark) alongside theological voices (Cobb, Neville, Panikkar, Levinas, Teilhard de Chardin), the paper proposes governance practices grounded in transparency, covenantal accountability, and inclusive deliberation. The result is a theologically informed agenda for policy and design that treats synthetic intelligences as potential participants in shared moral learning, while protecting human dignity and promoting ecological flourishing.

Throughout Christian history, technologies have shaped how individuals engage in spiritual reflection—from the codex and the printing press to contemporary digital media. Scholars of digital religion such as Heidi Campbell note that new technological environments often reshape religious practice and authority. Large language models may represent the next stage in this development.

This paper examines the emergence of AI-mediated tools designed to facilitate spiritual reflection through structured dialogue. Rather than delivering doctrinal instruction, these systems function as conversational partners that ask reflective questions, identify patterns across entries, and generate summaries that help individuals articulate what they believe, doubt, or are still exploring. Drawing on Michel Foucault’s concept of “technologies of the self,” the paper argues that AI may represent a new form of religious technology that assists individuals in examining and narrating their own spiritual lives.

By exploring AI as a reflective companion rather than a religious authority, this paper considers how such tools may reshape spiritual direction, faith exploration, and contemporary religious practice.