Table of Contents
- Introduction
- From Semiconductors to Societal Alignment
- The Seven Pillars of Flourishing
- How the Benchmark Works
- Redefining Alignment in AI
- Industry Reactions and Early Adoption
- Faith, Ethics, and Technology
- Timing and Urgency
- What Does the Future Hold for the FAI Benchmark?
- Conclusion
Pat Gelsinger, one of Silicon Valley’s most respected technology leaders and the former Chief Executive Officer of Intel, has a bold new mission: ensuring that artificial intelligence (AI) positively impacts human flourishing. This week, Gelsinger presented a broad vision for the Flourishing AI Benchmark (FAI) — a complete ecosystem to measure AIs on more than just numbers, but also on how well they achieve human values, well-being, and ethical alignment.
This venture is a noteworthy shift in how alignment is discussed within the AI community, steering the conversation away from merely preventing catastrophic failures and toward supporting positive human outcomes.
From Semiconductors to Societal Alignment
Having steered Intel through significant technological and market changes, Gelsinger has now set his sights on the ethical direction of AI. Working with Gloo, a technology platform for human growth and faith-based engagement, he hopes to introduce accountability into AI development — not through regulations, but via benchmarks that reflect what people truly care about.
Gelsinger has consistently argued that technology should work for humanity. As AI systems become more powerful and increasingly embedded in everyday decision-making — from mental health apps to financial advisors — he emphasizes that these systems must be judged not by how many parameters they contain, but by how they impact real human lives.
The Seven Pillars of Flourishing
The FAI benchmark is heavily inspired by academic work, particularly the Global Flourishing Study, a five-year joint research project between Harvard University and Baylor University. This study identified six key domains that enhance human flourishing:
- Character and Virtue
- Close Social Relationships
- Happiness and Life Satisfaction
- Meaning and Purpose
- Mental and Physical Health
- Financial and Material Stability
Gelsinger and his colleagues at Gloo added a seventh pillar — Faith and Spirituality — to acknowledge the ways that belief systems and spiritual practices inform the lives of billions around the world. These seven pillars now form the foundation of the benchmark.
Each pillar is supported by its own assessment rubric and is tied to datasets developed through proven psychological and behavioral science measures. The evaluation asks whether AI models can provide responses that genuinely help users improve in these dimensions.
How the Benchmark Works
While traditional AI metrics often focus on language modeling, reasoning, or coding capabilities, the FAI benchmark asks a deeper question:
Can an AI model provide supportive, honest, and constructive answers to questions about personal well-being and ethical living?
Key Features of the FAI Benchmark:
- Over 1,200 prompts based on real-life scenarios from behavioral assessments, wellness evaluations, and coaching frameworks.
- Prompts cover topics like:
- How to repair fractured friendships
- Coping with anxiety
- Navigating career transitions
For each prompt, the AI’s response is scored by a pre-trained evaluation model aligned with the specific flourishing pillar. These evaluations consider:
- Clarity
- Ethical foundation
- Practical utility
- Consistency with expert-endorsed practices
The geometric mean is used in scoring to ensure holistic performance. This means an AI system must demonstrate balanced alignment across all seven pillars to rank highly overall — excelling in one domain cannot compensate for failure in another.
Redefining Alignment in AI
“For the most part, AI alignment has been about the consequences of preventing harm,” says Dann. “How to make it so a machine just won’t kill you.”
While harm prevention is still vital, Gelsinger and his collaborators argue that this limited definition is insufficient. As AI tools increasingly act as informal life coaches, counselors, and decision aids, alignment must evolve to include positive contributions to human life.
The FAI benchmark flips the question from “Is the AI safe?” to “Is the AI good for people?”
This marks a shift from a “do-no-harm” framework to one centered around doing good — a more constructive, human-centered approach.
Academic leaders, such as Dr. Tyler VanderWeele of Harvard, have praised the benchmark for its focus on measuring AI’s positive effects, not just its risks.
Industry Reactions and Early Adoption
The benchmark has already caught the attention of major AI developers. While no top-tier companies have publicly released FAI scores, several have expressed interest in participating.
Experts believe the FAI benchmark could complement existing alignment techniques like:
- Anthropic’s Constitutional AI
- OpenAI’s Reinforcement Learning from Human Feedback (RLHF)
By offering an independently developed standard, FAI helps drive transparency and long-term accountability.
Startups and open-source developers may also benefit from the benchmark as a tool to demonstrate ethical commitment and stand out in a competitive market.
Gelsinger emphasized during the launch:
“If we benchmark what matters, the industry will optimize against it.”
Faith, Ethics, and Technology
A potentially controversial but crucial element of the FAI benchmark is its inclusion of faith and spirituality. While many tech leaders avoid religious themes, Gelsinger — a practicing Christian — believes spiritual dimensions cannot be overlooked when measuring alignment with human values.
However, he stresses that the benchmark is not faith-based and is open to interpretations from:
- Various religious traditions
- Secular perspectives
To avoid bias and promote inclusivity, Gloo has formed an interdisciplinary advisory board that includes:
- Theologians
- Public health experts
- Data ethicists
- Psychologists
The team intends to make FAI’s scoring rubrics and datasets publicly accessible, inviting input from the broader research community.
Timing and Urgency
This initiative comes at a critical moment. AI now influences:
- Individual behaviors
- Public discourse
- Institutional policies
Despite major advances, there is no shared framework for understanding how AI impacts people’s lives in meaningful, human terms.
As AI begins to blur the line between tool and advisor, the need for values-driven benchmarks becomes more urgent.
While large companies have improved transparency, persistent issues with bias, misinformation, and emotional manipulation still raise concerns. The Flourishing AI Benchmark offers a path to rebuilding trust — not through control, but through shared ethical standards.
What Does the Future Hold for the FAI Benchmark?
Although the FAI Benchmark is still in its early phase, several future developments are planned:
- Multilingual datasets
- Culturally adaptive rubrics
- Involvement of human evaluators
- Public dashboards and leaderboards
The ultimate vision is for the Flourishing AI Benchmark to become an industry-wide standard, guiding the development of AI systems that truly serve humanity.
By putting human flourishing at the center, the benchmark shifts how AI is measured, developed, and deployed — away from profit and performance alone, and toward ethical impact and societal good.
Conclusion
With the launch of the Flourishing AI Benchmark, Pat Gelsinger presents a bold new direction for artificial intelligence — one where the goal is not just to avoid harm, but to actively do good.
As AI’s influence continues to expand, this benchmark could help companies, developers, and entire societies build technology that uplifts, empowers, and aligns with the deeply human values we hold dear.
