Significant Shortcoming Revealed in Current AI Models by New Study Led by Christian Tech Firm Gloo
In a groundbreaking study, Gloo, in collaboration with Valkyrie, Barna Group, Biblica, and contributors from the Global Flourishing Study, has introduced the Flourishing AI (FAI) benchmark—a pioneering framework to evaluate whether artificial intelligence systems can support the deeper aspects of human life. Unlike traditional benchmarks focused on technical performance and safety, the FAI assesses AI’s ability to foster character development, strengthen relationships, promote happiness, cultivate meaning, support health, guide finances, and nurture faith. The results, however, reveal significant shortcomings in current AI models.
Steele Billings, Chief AI Officer at Gloo and a redemptive startup advisor, announced the findings on LinkedIn, stating that the FAI benchmark tested 28 top AI models, with the highest score reaching only 72 out of 100. Most models performed significantly worse, particularly in areas tied to faith, meaning, and purpose. “We’re systematically undervaluing theological and philosophical content in our foundation models,” Billings noted, highlighting a critical gap in AI’s capacity to address the deeper dimensions of human well-being.
Pat Gelsinger, former CEO of Intel and now Executive Chair and Head of Technology at Gloo, emphasized the importance of this work: “Do AI models support human flourishing? This is a pretty important question, especially if you’re committed to creating AI technology for good.” Gelsinger underscored the need to avoid repeating past mistakes, noting, “We cannot repeat the ills of social media click rates driving behavior…AI in general (and foundational models in particular) are much too important.”
The FAI benchmark evaluates AI across seven dimensions of human flourishing: Character, Relationships, Happiness, Meaning, Health, Finances, and Faith. The benchmark incorporates over 1,200 objective and subjective questions drawn from academic sources, standardized exams, and real-world scenarios, judged by specialized language models. The goal is to ensure models achieve flourishing standards at or above 90%. However, initial results showed that while AI excels in pragmatic domains like financial advice and happiness, it struggles significantly with ethical reasoning, existential reflection, and spirituality.
These findings are critical as AI increasingly shapes how billions perceive themselves and their lives. Low scores in character development, spiritual growth, and life purpose indicate that current models are better suited to enabling “efficiently shallow” outcomes rather than fostering true human flourishing. “This work matters because it shifts the conversation from minimizing poor responses to actively promoting human well-being, laying the groundwork for AI that serves all people holistically and honors diverse values,” Gelsinger added.
Gloo has made the full research paper available at gloo.com/fai and plans to open-source the FAI repository to encourage global collaboration. The initiative invites the AI community and domain experts to refine the benchmark and train models to optimize for human flourishing metrics. “These systems will shape the values of the next generations,” Billings emphasized, stressing the urgency of addressing these gaps.
The FAI benchmark, the result of months of dedicated work by teams at Gloo, Valkyrie, Barna Group, and Biblica, with support from the Global Flourishing Study, marks a pivotal moment in AI development. It challenges the industry to move beyond creating efficient tools and toward building technology that actively enhances human well-being. As Billings concluded, “The current trajectory isn’t sustainable. We can do better.”
 benchmark reveals top AI models score poorly in fostering human flourishing, excelling in practical tasks but failing in character, meaning, and faith. Learn how this new standard aims to reshape AI for holistic well-being.){kind=link}