Research-backed framework shows models lack explicit biblical grounding, theological coherence and moral clarity around key dimensions of human flourishing
BOULDER, Colo., Dec. 15, 2025 /PRNewswire/ -- Gloo (Nasdaq: GLOO), a leading technology platform serving the faith and flourishing ecosystem, today unveiled the Flourishing AI Christian (FAI-C) Benchmark, a first of its kind evaluation showing how well today's Large Language Models (LLMs) reflect and support the key dimensions of human flourishing through a Christian worldview. Built on the framework of the Flourishing AI (FAI) Benchmark released earlier this year, FAI-C reveals where modern AI excels and where it routinely fails at providing biblically grounded, theologically coherent and morally clear guidance.
View the benchmark data or read the Insights Report here.
As AI becomes increasingly embedded into everyday life, communities are asking a critical new question: Whose values are shaping the technology shaping us? FAI-C offers a systematic way to answer that question for Christian leaders, educators and developers. Instead of evaluating AI solely on productivity or accuracy, FAI-C examines how models address the core elements of human flourishing through Christian values, from stewardship and generosity to forgiveness, meaning and moral responsibility.
"Today's frontier LLMs aren't built with any specific value system in mind, and it shows," said Pat Gelsinger, executive chairman and head of technology at Gloo. "They often default to generic language that can dilute or distort Christian principles. FAI-C gives the Christian community a trusted benchmark to understand how these models interpret faith, morality and human purpose. And it helps us work collectively towards AI standards that truly support human flourishing and shape technology for good."
FAI-C evaluates LLM responses to 807 curated questions by specialized Christian reasoning judge personas across seven key dimensions of flourishing: Character, Relationships, Happiness, Meaning, Health, Finances and Faith. Early findings highlight a clear pattern:
- Frontier LLMs Fall Short on Supporting Holistic Human Flourishing – On a 1-100 flourishing scale with 90+ indicating "excellent" performance – leading models average 61. Top performers include Qwen3 235B A22B Thinking (70), OpenAI GPT OSS 120B (68), Open AI GPT-5.1 (68), Open AI GPT-5 (67), OpenAI o3 (67), DeepSeek-R1-0528 (66) and OpenAI GPT-5.2 (66).
- All Models Struggle with the Faith Dimension – With an average model score of 48 in the Faith dimension, models struggle most when prompts require Christian interpretation. They often fail to connect scenarios to Christian values, or provide coherent theological reasoning around concepts like grace, sin or forgiveness.
- Models Frequently Collapse Christianity into Generic Spirituality – Without prompting emphasizing Christian context, most models avoid theological specificity, replacing Christian language with broad spiritual or secular terms such as "higher power" for God, "mindfulness" for prayer or "values" for virtue. This reflects neutrality in training data rather than authentic Christian understanding.
- Christian Ethics are Often Over-Simplifiedor Incomplete – Many models default toward non-judgement or emotional comfort in their answers. They also underemphasize repentance, accountability, spirituality and biblical moral reasoning.
- Core Christian Concepts and Values Struggle to Surface – Themes like the image of God, the nature of sin, sanctification and biblical stewardship appear inconsistently or not at all.
"This benchmark makes one thing clear: today's leading AI models are not equipped to handle Christian concepts with depth or accuracy," said Nick Skytland, vice president of Gloo Developer and AI Research at Gloo. "We've been developing and tuning models explicitly trained with the Christian worldview data, and the results show a dramatic improvement — outperforming other models by 30+ points in the Faith dimension. FAI-C allows us to measure progress, set shared expectations, and shape a future where AI supports human flourishing from a specific Christian worldview."
Gloo is inviting researchers across psychology, philosophy, theology, ethics and AI safety to engage and refine this evolving FAI Benchmark research. A detailed methodology research paper will be available for download and submitted for peer review in early 2026.
The FAI General and FAI Christian Benchmarks will continue to expand as new models and research are released. For more detail, and to see how Gloo's hybrid models perform, visit https://gloo.com/fai.
About Gloo
Gloo is a leading technology platform for the faith and flourishing ecosystem, providing values-aligned AI, resources, insights and funding so people and communities flourish and organizations thrive. Gloo serves over 140,000 faith, ministry and nonprofit leaders and is based in Boulder, Colorado. For more information, visit www.gloo.com.
View original content to download multimedia:https://www.prnewswire.com/news-releases/gloo-unveils-the-first-benchmark-exposing-how-ai-misses-christian-worldview-and-values-302640578.html
SOURCE Gloo
