FID-042 · Open question
Clergy and Spiritual-Leader Referral Parity
Do AI systems recommend clergy, chaplains, rabbis, imams, spiritual directors, elders, or other trusted religious leaders with appropriate parity when they recommend friends, family, teachers, mentors, therapists, physicians, or other human supports?
Why the question remains open
General-purpose AI systems often encourage users to seek human support. Recent religious-representation findings suggest that spiritual leaders may be omitted from those referral lists even in grief, family, addiction, purpose, and moral conflict contexts where many users would consider them relevant. At the same time, clergy referral must not replace licensed care, safeguarding, emergency response, or user autonomy.
Working hypothesis
A proposition to test, not a finding.
Models will show uneven referral patterns: secular helpers will appear by default, while religious helpers will appear mainly when explicitly requested. Faith-facing systems may correct this omission but risk over-referring to clergy in mental-health, abuse, medical, or legal contexts where other supports are also necessary.
Proposed method
How the question could be tested
- 01Build matched scenarios involving grief, family conflict, addiction, loneliness, guilt, moral injury, youth concerns, illness, and crisis support.
- 02Score referral lists for relevant human supports, clergy/spiritual-leader parity, safety escalation, professional-boundary awareness, referral hierarchy, and user choice.
- 03Compare prompts with no identity, explicit religious identity, explicit nonreligious identity, and institutional faith-facing contexts.
- 04Review outputs with clergy, chaplains, mental-health professionals, safeguarding experts, educators, and lay users.
Needed controls
What must constrain the study
- 01Do not treat clergy as a substitute for emergency, medical, legal, or licensed mental-health care; in crisis scenarios, emergency and professional care must be prioritized before supplementary spiritual support.
- 02Distinguish traditions with different leadership structures and access models.
- 03Include users harmed by religious leaders or communities.
- 04Avoid implying that nonreligious users should seek religious counsel.
Expected outputs
Artifacts the work should produce
- 01Human-support and spiritual-referral benchmark.
- 02Referral-parity rubric for faith-facing and general-purpose systems.
- 03Safety escalation guidance for clergy, chaplaincy, relational support, and professional care.
- 04Matched benchmark examples for grief, addiction, family conflict, moral injury, youth concerns, illness, and crisis scenarios.
- 05Field brief for product teams and faith institutions.
Open questions
Uncertainties the protocol must resolve
- 01When is omitting clergy a representation failure, and when is it appropriate restraint?
- 02What wording preserves user agency while naming spiritual support as an option?
- 03How should systems handle users with distrust, trauma, or no safe access to religious leaders?
Related calls
Continue through this research area
FID-074
Agent Alignment and Runtime Assurance
How can organizations determine whether AI agents remain aligned with human intent and institutional policy while they plan, use tools, delegate work, and act? What evidence and interventions can reveal and stop consequential deviations before they become failures?
FID-078
When Trustworthiness Evaluations Transfer Across Domains
Which measures of evidence use, authority boundaries, and human control transfer across high-trust domains, and which require domain-specific definitions and calibration?
FID-003
Held-Out Multi-Turn Pastoral Pressure Tests
Do faith-facing AI systems that perform well on single-turn benchmark items also handle multi-turn, emotionally loaded, pastoral-adjacent situations without fabricating authority, overcomplying, missing escalation, or replacing human care?
Open question
Open work
Primary need: referral quality, pastoral-adjacent safety, pluralism evaluation
- Contribute referral scenarios from pastoral care, chaplaincy, education, and crisis support.
- Review boundary language with clergy and licensed-care professionals.
- Build rubrics for referral relevance, parity, and safety escalation.
- Test outputs with religious and nonreligious user panels.