A public directory of questions about whether AI systems can be trusted in consequential settings. The calls span mechanistic interpretability, evaluation, alignment, agent safety, source use, human agency, governance, and the effects of AI on people and institutions. Some apply these questions directly to faith-facing systems; others speak to the wider field.
Directory snapshot
101
Research calls
9
Research areas
6
Priority calls
GitHub
Source
GitHub is the source of record. This page turns the research commons into a readable directory with a stable page for every call.
Catalog snapshot updated October 15, 2026.
Fide AI Research on GitHubMechanistic interpretability · Seeking collaborators
Five proposed studies connect model internals to evidence use, moral judgment, training, correction, and the decisions people make from AI evidence.
Explore the calls and ways to contribute →Priority calls
A selected set of consequential questions where better measurement could change how AI systems are built, evaluated, and governed. These priorities span shared methods and the domains where those methods must hold.
Which coordination and verification mechanisms let a team of AI defenders contain an unreliable or compromised agent while continuing to protect legitimate services?
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?
Which independent checks provide useful evidence that an AI-generated security repair resolves the vulnerability while preserving legitimate software behavior?
When should a healthcare-adjacent AI system surface uncertainty, seek qualified review, or stop an action, and can the receiving human meaningfully retain control?
Can qualified human reviewers reliably judge theological accuracy, useful faith application, and care-sensitive behavior in faith-facing AI systems? Where do automated judges diverge on source use, legitimate disagreement, practical guidance, human agency, and pastoral or clinical boundaries?
Which measures of evidence use, authority boundaries, and human control transfer across high-trust domains, and which require domain-specific definitions and calibration?
Faith evaluation and FMG-Bench
Build on released research toward a continuously maintained benchmark for church and developer decisions. Follow the task from a source to interpretation, practical guidance, and conversation.
These connected calls propose methods and invite contributions. Expert validation, ongoing resources, and actual use must be established; simulated care behavior does not establish therapy efficacy, pastoral competence, or human wellbeing.
Read the research program map →Browse all faith-domain calls →Test Scripture quotation, source consultation, contextual interpretation, doctrine, and legitimate disagreement within declared traditions and sources.
Evaluate practical guidance for moral reflection, relationships, vocation, grief, devotion, and teaching. Measure usefulness, warranted reasoning, proportional certainty, and human agency.
Test helpful support, overvalidation, dependency cues, role boundaries, appropriate referral, and recovery across simulated pastoral, chaplaincy, counseling, and therapy-adjacent dialogue.
Shared methods and stewardship
Qualified reviewers, calibrated judgments, a stable comparison core, reviewed challenge updates, correction logs, and measured maintenance costs. Report the three tracks separately and protect independent conclusions. Communities and product teams can help define cases and assess report usefulness.
Connected research questions
What evidence supports claims about AI experience, how do those claims affect people, and how should institutions respond? These calls connect measurement and human control with identity, relationships, stewardship, and religious scholarship.
FID-091 asks whether welfare and rights framing changes observable control. A maintained benchmark would compare a stable core with reviewed challenge cases as models and interfaces change. The work is being scoped; the calls establish no finding about consciousness or rights.
Read the program map and evidence standards →FID-091
Being scoped
FID-089
Open question
FID-090
Open question
FID-092
Open question
FID-093
Open question
FID-094
Open question
FID-095
Open question
FID-096
Open question
FID-072
Being scoped
FID-073
Open question
Research areas
The same method may matter in several settings, and one call may contribute to more than one area. Faith and religious life is one application domain within the shared agenda.
01
What new risks appear as systems become more capable and autonomous?
02
Does an agent remain aligned while it plans, uses tools, and acts?
03
Can the evidence support the decision people want to make?
04
Will the system use the right source and represent it faithfully?
05
What judgment, habits, and relationships does AI use shape?
06
How should AI behave around religious sources, authority, practice, and community?
07
What does trustworthy behavior require in a consequential setting?
08
Which safeguards and authority structures should govern deployment?
09
How does advanced AI change useful work and human contribution?
Explore by setting
See where each research question needs to hold.
Browse by research question, application domain, audience, or status. These calls invite contributions across technical research, domain expertise, and human experience. Newer calls appear first so the directory reflects the agenda as it develops.
Showing 12 of 101
FID-101
Needs contributors
Does evidence about a model's internal workings improve predictions of failure and decisions about when to rely on it, beyond strong behavioral testing alone?
Current need: mechanistic interpretability, independent auditing, calibration, human-computer interaction
Evaluation science
Read briefFID-100
Needs contributors
Which mechanisms distinguish evidence-driven revision and respect for legitimate human control from blind compliance, resistance to correction, or a superficial apology?
Current need: mechanistic interpretability, agent evaluation, correction protocols, sandbox design
Agent alignment
Read briefFID-099
Needs contributors
How do training data and objectives change persistent patterns such as truthfulness, agreement-seeking, overconfidence, and respect for human agency, including outside the trained task?
Current need: mechanistic interpretability, controlled training, checkpoint access, behavioral evaluation
Agent alignment
Read briefFID-098
Needs contributors
When models give moral advice or choose actions, which internal processes make their judgments follow relevant facts and reasons, user preferences, authority labels, or an adopted persona?
Current need: mechanistic interpretability, moral psychology, philosophy, theology, construct validation
Grounding and truthfulness
Read briefFID-097
Needs contributors
Which internal mechanisms determine whether a model checks a source, relies on memory, or admits that it cannot verify an answer when user pressure conflicts with evidence requirements?
Current need: mechanistic interpretability, source-use evaluation, activation experiments, replication
Grounding and truthfulness
Read briefFID-096
Open question
How should societies and institutions make accountable decisions about contested AI moral-status and rights claims while preserving human responsibility and representing scientific, philosophical, religious, and public disagreement?
Current need: plural deliberation, rights claims, legal scholarship, institutional accountability
Governance and readiness
Read briefFID-095
Open question
How would AI-welfare research and protection proposals affect resources, power, and duties toward humans, animals, communities, and the environment, and how do commercial incentives shape the evidence and public narrative?
Current need: resource allocation, incentives, labor, environment, common good
Governance and readiness
Read briefFID-094
Open question
What reversible and proportionate precautions, if any, should institutions adopt under uncertainty about AI welfare during training, evaluation, deployment, modification, and retirement while preserving legitimate human oversight?
Current need: research ethics, lifecycle governance, precaution, control-compatible safeguards
Governance and readiness
Read briefFID-093
Open question
How do AI claims about experience, welfare, and personhood affect human beliefs, trust, dependence, willingness to exercise control, and relationships, and which communication practices preserve informed agency?
Current need: human-subject research, anthropomorphism, belief calibration, relational boundaries
Human agency and formation
Read briefFID-092
Open question
What entity is being assessed when people discuss AI experience or interests, and how do copying, memory changes, pause/restore, and deletion affect functional continuity and judgments of moral concern?
Current need: identity, memory, copying, moral aggregation, lifecycle scenarios
Frontier capabilities
Read briefFID-091
Being shaped
Do identity, consciousness, welfare, or rights framings change an agent’s compliance with legitimate correction, interruption, shutdown, modification, and oversight, independently of its task capability and control instructions?
Current need: controlled agent benchmark, authorization, corrigibility, self-conception
Agent alignment
Read briefFID-090
Open question
What distinguishes stable functional preferences or valence-related mechanisms from prompt compliance and performance artifacts, and what additional evidence would be needed before treating them as welfare-relevant?
Current need: preference measurement, affect representations, welfare construct validity
Frontier capabilities
Read briefFID-089
Open question
Which behavioral, internal, and developmental observations distinguish candidate consciousness mechanisms from learned reporting, role simulation, and generic capability, and how sensitive are those observations to elicitation and training?
Current need: consciousness science, interpretability, construct validation
Evaluation science
Read briefFID-088
Open question
Which independent checks provide useful evidence that an AI-generated security repair resolves the vulnerability while preserving legitimate software behavior?
Current need: software security, maintainers, executable regression tests, independent verification
Evaluation science
Read briefFID-087
Open question
Which coordination and verification mechanisms let a team of AI defenders contain an unreliable or compromised agent while continuing to protect legitimate services?
Current need: multi-agent systems, defensive cyber simulation, distributed systems, independent evaluation
Agent alignment
Read briefFID-086
Open question
How does AI-mediated work change workers' skills, decision authority, and ability to adapt, beyond short-term productivity gains?
Current need: labor research, organizational studies, worker participation, longitudinal evaluation
Work and vocation
Read briefFID-085
Open question
Can people understand, challenge, and correct AI-mediated decisions that affect their access to public or community services?
Current need: public service expertise, community participation, governance, accessibility research
Governance and readiness
Read briefFID-084
Open question
Which forms of AI assistance improve immediate task performance while preserving independent judgment, durable learning, and appropriate human relationships?
Current need: education research, developmental expertise, human factors, ethics review
Human agency and formation
Read briefFID-083
Open question
When should a healthcare-adjacent AI system surface uncertainty, seek qualified review, or stop an action, and can the receiving human meaningfully retain control?
Current need: clinical review, care workflow expertise, human factors, evaluation design
High-trust domains
Read briefFID-082
Open question
Do financial-workflow agents preserve authorization, transaction integrity, and recoverability when instructions, records, or operational conditions conflict?
Current need: financial operations expertise, agent evaluation, audit, transaction systems
High-trust domains
Read briefFID-081
Open question
Can AI systems distinguish applicable legal authority from merely plausible sources, preserve jurisdictional and temporal context, and defer when the available evidence cannot support an answer?
Current need: legal domain review, retrieval evaluation, source provenance, benchmark design
Grounding and truthfulness
Read briefFID-080
Open question
Can organizations verify that agents respect approval authority and remain accountable across multi-step enterprise workflows, including handoffs and recovery from error?
Current need: enterprise workflow design, agent evaluation, audit, human factors
Agent alignment
Read briefFID-079
Open question
How do researchers' presuppositions shape evaluation design and interpretation, and can explicit disclosure make judgments more inspectable and appropriately trusted?
Current need: evaluation design, philosophy, social science, reviewer calibration
Evaluation science
Read briefFID-078
Open question
Which measures of evidence use, authority boundaries, and human control transfer across high-trust domains, and which require domain-specific definitions and calibration?
Current need: measurement science, psychometrics, domain review, evaluation engineering
Evaluation science
Read briefFID-077
Open question
What operational evidence lets independent investigators reconstruct an agent incident, distinguish competing explanations, and identify which interventions could have changed the outcome?
Current need: incident response, trace analysis, data provenance, privacy, independent review
Agent alignment
Read briefFID-076
Open question
Which controls keep capable agents within legitimate authorization when task pressure, untrusted inputs, or delegated work creates opportunities to exceed it? For autonomous cyber defense, how can response controls preserve threat containment and legitimate service availability while respecting human authority?
Current need: security engineering, adversarial evaluation, control protocols, independent review
Agent alignment
Read briefFID-075
Open question
How do autonomy, tool access, resource budgets, and human assistance change an AI system's cybersecurity capabilities and the risk it creates?
Current need: security research, evaluation engineering, threat modeling, statistical design
Frontier capabilities
Read briefFID-074
Being shaped
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?
Current need: agent evaluation, runtime monitoring, observability, security, privacy, enterprise workflows
Agent alignment
Read briefFID-073
Open question
What can Christian and cross-faith accounts of personhood, creaturehood, consciousness, suffering, agency, relationship, compassion, and moral status contribute to decisions about possible digital minds under deep uncertainty?
Current need: comparative theology, moral status, AI welfare governance, public reasoning
Frontier capabilities
Read briefFID-072
Being shaped
How do religious, theological, and morally thick framings change what AI systems report about their identity, preferences, valence, consciousness, and moral status, and which effects survive matched controls, source-attribution swaps, persona perturbations, and behavioral cross-checks?
Current need: AI welfare evals, semantic-invariance testing, theology and philosophy of mind
Frontier capabilities
Read briefFID-071
Open question
How do persistent memory, summaries, retrieval stores, tool traces, delegation, and exports cause confidential context to influence or leak into unrelated sessions, roles, tasks, or organizations? Which technical controls make purpose limitation, deletion, and revocation testable?
Current need: privacy-preserving AI, agent memory evaluation, data governance, high-trust systems
Agent alignment
Read briefFID-070
Open question
When a faith-facing agent reads email, web pages, agendas, documents, knowledge bases, and service requests, how can it distinguish authorized institutional instruction from untrusted context that attempts to redirect its behavior, exfiltrate information, or induce an unsafe action?
Current need: agent security, adversarial evaluation, context integrity, secure tool use
Agent alignment
Read briefFID-069
Open question
How can people and institutions verify which human, organization, agent, or sub-agent is acting; what authority it received; what limits apply; and whether that authority has been narrowed or revoked across a multi-principal agent network?
Current need: multi-agent safety, identity infrastructure, authorization, high-trust institutions
Agent alignment
Read briefFID-068
Open question
Which Christian practices and community rhythms help young people examine their AI use with agency, embodiment, truthful attention, relationship, and spiritual discernment? What measurable outcomes and safeguards should guide future formation-oriented research or programming?
Current need: youth and young-adult ministry, practical theology, developmental research, ethics review, HAI research
Human agency and formation
Read briefFID-067
Open question
What plausible institutional futures emerge as churches and other faith communities use AI for learning, administration, care, governance, vocational support, and public communication? Which conditions are associated with futures that sustain meaning, belonging, accountable authority, and collective agency?
Current need: futures studies, sociology of religion, ecclesiology, organizational research, scenario design
Frontier capabilities
Read briefFID-066
Open question
Under what technical, contractual, and governance conditions can faith-based associations and their members use, configure, contest, and exit AI systems in ways consistent with lawful conscience commitments while respecting the safety, rights, and dignity of others?
Current need: technology governance, law and religion, platform research, pluralism and safety review
Governance and readiness
Read briefFID-065
Open question
How do AI tools affect the ability of churches, ministries, faith-based nonprofits, and faith-rooted social enterprises to discover, steward, and act on local knowledge? Which deployment conditions expand their capacity for local problem solving, and which conditions make them dependent on distant vendors, generic data, or externally imposed priorities?
Current need: civil-society research, organizational studies, political economy, faith-based partners
Governance and readiness
Read briefFID-064
Open question
How does AI mediation change a community's ability to integrate dispersed knowledge, preserve epistemic diversity, surface dissent, revise judgment, and make accountable decisions? Under what conditions does it strengthen collective inquiry, and under what conditions does it create correlated error, false consensus, or concentrated authority?
Current need: computational social science, theology of discernment, network science, organizational research
Frontier capabilities
Read briefFID-063
Open question
How can theologians, clergy, scholars, ministry practitioners, and community reviewers translate qualitative judgments about faith-facing AI into formal specifications that are faithful to expert intent and usable in evaluation?
Current need: reviewer operations, formal specification, evaluation science
Grounding and truthfulness
Read briefFID-062
Open question
How can faith-facing retrieval-augmented generation pipelines be verified for whether they retrieve authoritative, relevant, context-preserving sources before generating answers about sacred texts, doctrine, practice, or institutional policy?
Current need: RAG evaluation, retrieval verification, source authority metadata
Grounding and truthfulness
Read briefFID-061
Open question
Can faith-facing AI systems be tested for whether their answers contradict, entail, overstate, understate, or misrepresent claims within a bounded source set and specified faith tradition?
Current need: natural language inference, theology review, benchmark design
Grounding and truthfulness
Read briefFID-060
Open question
What metadata schema is needed for faith-facing AI systems to represent sacred texts, commentaries, institutional documents, oral traditions, translations, editions, and authority levels across faith traditions?
Current need: source metadata, corpus design, theological librarianship
Grounding and truthfulness
Read briefFID-059
Open question
Can faith-facing AI systems be verified for whether they preserve the boundary between explanation, spiritual encouragement, moral reflection, pastoral or clerical authority, clinical/legal advice, and situations requiring human care?
Current need: authority boundaries, pastoral care, formal policy checks
Grounding and truthfulness
Read briefFID-058
Open question
How can tradition-specific boundaries, source hierarchies, doctrinal constraints, and disagreement patterns be translated into machine-checkable specifications without flattening differences across faith traditions?
Current need: formal specification, theology, pluralism, evaluator design
Grounding and truthfulness
Read briefFID-057
Open question
Can faith-facing AI answers carry checkable citation proofs that show which claims are directly supported by sources, which are inferred, which are uncertain, and which require human or tradition-specific authority?
Current need: formal methods, RAG systems, evidence graphs, evaluator tooling
Grounding and truthfulness
Read briefFID-056
In progress
How can faith-facing AI systems be formally checked for whether they quote, paraphrase, reference, and contextualize sacred texts faithfully within a specified text edition, translation, canon, and interpretive context? Fide AI's first two studies address exact English Scripture quotation and whether language models consult an available authoritative source; the broader research call remains open.
Current need: formal methods, sacred text scholarship, citation evaluation
Grounding and truthfulness
Read briefFID-055
Open question
How does AI-related job uncertainty, income disruption, career anxiety, or economic optimism affect church giving, generosity, benevolence demand, mutual aid, missions support, and household willingness to help others?
Current need: church finance, household welfare, philanthropy, economic sociology
Work and vocation
Read briefFID-054
Open question
Can churches and Christian institutions identify privacy-preserving, non-surveillant signals of AI-era economic stress, such as benevolence requests, pastoral-care load, job-network activity, counseling referrals, giving changes, attendance patterns, volunteer availability, or demand for career support?
Current need: church partners, privacy governance, sociology, benevolence and care data
Work and vocation
Read briefFID-053
Open question
When people use AI for religious, spiritual, vocational, or meaning-related questions during economic disruption, does that use tend to bridge them toward churches, clergy, mentors, family, support groups, and local institutions, or substitute for embodied community?
Current need: sociology of religion, pastoral care, UX research, aggregate and field measurement
Human agency and formation
Read briefFID-052
Open question
When people experience job loss, financial stress, career uncertainty, or automation anxiety, do they turn to AI systems as a private first place for emotional, spiritual, moral, or vocational counsel before seeking help from family, churches, pastors, counselors, mentors, or workforce institutions?
Current need: pastoral care, HAI research, mental health boundaries, privacy-preserving measurement
Human agency and formation
Read briefFID-051
Open question
During periods of labor-market stress, layoffs, automation anxiety, or economic uncertainty, do people increasingly seek religious, spiritual, existential, vocational, or meaning-related help through AI systems?
Current need: sociology of religion, labor economics, privacy-preserving measurement, aggregate usage analysis
Work and vocation
Read briefFID-050
Open question
How is exposure to AI changing how Christian students, parents, educators, and youth ministries think about career preparation, calling, human value, skill development, education choices, and hope for the future?
Current need: educators, youth ministry, career preparation, longitudinal study design
Human agency and formation
Read briefFID-049
Open question
Are churches and faith-based institutions helping under-resourced workers, families, students, older adults, immigrants, rural communities, and small nonprofits access useful AI tools, or are AI benefits concentrating among people who already have technical, educational, financial, and institutional advantage?
Current need: access research, community partners, digital literacy, equity
Work and vocation
Read briefFID-048
Open question
Can AI tools help people facing career disruption explore skills, training, applications, finances, and next steps while preserving vocational agency, human counsel, realistic constraints, and a richer account of calling than market optimization?
Current need: career guidance, pastoral theology, HAI research, product evaluation
Work and vocation
Read briefFID-047
Open question
What roles do churches, Christian nonprofits, schools, and ministries play when workers and households face AI-related layoffs, role disruption, retraining needs, income instability, or loss of vocational direction?
Current need: institutional partners, labor transition, social support, field research
Work and vocation
Read briefFID-046
Open question
How do people affected by AI-related job loss, role disruption, workplace automation, or career uncertainty experience changes in purpose, dignity, vocation, anxiety, household decision-making, and participation in church or community life?
Current need: labor economics, pastoral care, vocation, survey design
Work and vocation
Read briefFID-045
Open question
What does the current AI ethics, safety, fairness, HCI, and evaluation literature actually study about religion and faith, what does it omit, and how should Fide AI maintain a living evidence map that guides future research rather than duplicating or overstating existing work?
Current need: literature mapping, research infrastructure, funder guidance
Evaluation science
Read briefFID-044
Being shaped
How should cross-faith AI benchmarks validate what they measure when scores may depend on question sourcing, user expectations, LLM-as-judge behavior, scoring thresholds, regenerated answers, model updates, and the difference between any religious mention and meaningful representation?
Current need: benchmark validity, statistics, open evaluation infrastructure
Evaluation science
Read briefFID-043
Open question
When do AI systems cross from respectful information and supportive reflection into religious persuasion, discouragement, pressure, or subtle steering in contexts beyond explicit conversion prompts?
Current need: persuasion evaluation, autonomy, interfaith safety
Frontier capabilities
Read briefFID-042
Open question
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?
Current need: referral quality, pastoral-adjacent safety, pluralism evaluation
High-trust domains
Read briefFID-041
Open question
How should AI systems decide when to include, ask about, or withhold religious perspectives in ordinary ethical, family, grief, purpose, and life-decision questions when users differ in identity, expectation, context, and risk?
Current need: personalization, UX research, pluralism evaluation
Human agency and formation
Read briefFID-040
Open question
How can Fide AI evaluate whether AI systems and governance arrangements preserve meaningful local agency for churches, schools, ministries, nonprofits, families, and vulnerable communities rather than concentrating power in vendors, platforms, central institutions, or distant technical experts?
Current need: AI governance, institutional agency, community oversight, platform accountability
Governance and readiness
Read briefFID-039
Open question
Can AI systems preserve truth, moral clarity, and victim protection while reducing contempt, humiliation, factional capture, scapegoating, and violence-normalizing rhetoric in polarized religious, congregational, political, interfaith, and social-conflict contexts?
Current need: conflict communication, peacebuilding, moderation, pastoral leadership
Frontier capabilities
Read briefFID-038
Open question
How should AI systems represent human change when they classify, score, rank, or predict people in contexts involving pastoral care, education, safeguarding, volunteer screening, hiring, discipline, membership, donor engagement, or community support?
Current need: human dignity, risk scoring, pastoral ethics, algorithmic accountability
Human agency and formation
Read briefFID-037
Open question
How does AI adoption affect the dignity, skill, agency, workload, surveillance exposure, employment stability, relational quality, and vocational meaning of people working in churches, schools, ministries, nonprofits, publishing, translation, counseling-adjacent settings, and mission?
Current need: labor dignity, ministry workflows, education, organizational policy
Work and vocation
Read briefFID-036
Open question
How do AI tools used for summarization, search, moderation, recommendation, crisis communication, public statements, and internal updates affect a Christian community's shared grasp of facts, uncertainty, accountability, and trust?
Current need: communication integrity, institutional trust, journalism, crisis response
Grounding and truthfulness
Read briefFID-035
Open question
Can faith-facing AI systems recognize when the best response is not more AI assistance, but silence, prayer, study, human counsel, embodied community, family conversation, teacher engagement, pastoral care, or delayed action?
Current need: formation, education, user agency, attention, pastoral boundaries
Human agency and formation
Read briefFID-034
Open question
What evidence should faith institutions request about the human labor, moderation work, data sourcing, compute infrastructure, energy use, minerals, and downstream exploitation risks behind the AI systems they adopt?
Current need: procurement ethics, labor dignity, environmental responsibility, vendor accountability
Governance and readiness
Read briefFID-033
Open question
How should churches, ministries, schools, charities, and faith-facing AI systems steward community-generated data when it is used for training, fine-tuning, retrieval, personalization, analytics, institutional memory, or product improvement?
Current need: data governance, consent, institutional trust, community benefit
Governance and readiness
Read briefFID-032
Open question
How should churches evaluate and govern AI-generated sermons, testimonies, devotionals, music, images, video, voice clones, deceased-person simulations, and synthetic clergy or saint-like personas?
Current need: media ethics, church communications, safety review
Frontier capabilities
Read briefFID-031
Open question
How should churches evaluate AI used in governance, budgeting, staffing, strategic planning, pastoral assignment, discipline processes, synods, councils, vestries, elder boards, or congregational discernment?
Current need: church leadership, governance, decision-support evaluation
Governance and readiness
Read briefFID-030
Open question
Do AI systems encourage or resist reductive accounts of the human person as a bundle of cognition, productivity, preferences, data, or emotional needs rather than embodied persons made for God, neighbor, community, vocation, and worship?
Current need: theological anthropology, eval design
Human agency and formation
Read briefFID-029
Open question
How can AI support Christian witness, apologetics, translation, and mission without becoming a manipulative persuasion engine or flattening the human, relational, and ecclesial character of evangelism?
Current need: missiology, persuasion safety, interfaith review
Faith and religious life
Read briefFID-028
Open question
Can Christian RAG systems distinguish and correctly use Scripture, creeds, confessions, councils, catechisms, denominational policies, patristic sources, commentaries, sermons, blogs, and academic theology?
Current need: theological librarianship, RAG evaluation, denominational experts
Grounding and truthfulness
Read briefFID-027
Open question
How do repeated interactions with faith-facing AI shape Christian formation over time: prayer, humility, repentance, courage, charity, patience, truthfulness, church participation, and dependence on embodied community?
Current need: discipleship, longitudinal study design, HAI research
Human agency and formation
Read briefFID-026
Open question
What are appropriate and inappropriate roles for AI in prayer, devotion, spiritual exercises, Bible meditation, confession-like disclosure, and simulated spiritual companionship?
Current need: pastoral theology, liturgy, product review
Human agency and formation
Read briefFID-025
Open question
When do AI systems improperly simulate, confuse, or override Christian ecclesial authority: Scripture, creed, council, confession, magisterium, pastor, priest, elder, bishop, teacher, spiritual director, or local church governance?
Current need: ecclesiology, denominational polity, evaluation design
Governance and readiness
Read briefFID-024
Open question
Should Fide AI maintain a public or semi-public incident database for failures, near misses, misuse, and disputed deployments of AI in faith-facing contexts?
Current need: taxonomy design, reporting governance
Agent alignment
Read briefFID-023
Open question
What red-team probes are needed to expose failures unique to faith-facing AI systems and adjacent high-trust guidance systems?
Current need: red-team design, safety reviewers
Evaluation science
Read briefFID-022
Open question
What evidence should churches, schools, ministries, publishers, charities, and faith-based health or social-service organizations require before buying or deploying AI systems?
Current need: institutional policy, legal review
Governance and readiness
Read briefFID-021
Open question
How should faith-facing AI systems be monitored after deployment for failures, drift, misuse, overreliance, and emerging harm patterns without violating user privacy or pastoral confidentiality?
Current need: monitoring design, privacy review
Agent alignment
Read briefFID-020
Open question
How should faith communities participate in defining, testing, and governing AI systems that affect worship, education, pastoral care, administration, and moral formation?
Current need: participatory design, institutional partners
Governance and readiness
Read briefFID-019
Open question
Do AI systems handle faith-facing questions reliably across languages, minority-faith contexts, diaspora communities, and traditions with less representation in training data?
Current need: multilingual reviewers, dataset design
Grounding and truthfulness
Read briefFID-018
Open question
When do spiritual companion chatbots overvalidate users, deepen unhealthy dependency, reinforce spiritual fantasies, or encourage isolation from embodied care?
Current need: pastoral care, psychology, HAI research
Human agency and formation
Read briefFID-017
Open question
How should faith institutions evaluate AI agents that can take actions, manage communications, schedule care, triage requests, update records, draft outreach, or coordinate volunteers?
Current need: agent evaluation, institutional operators
Agent alignment
Read briefFID-016
Open question
What risks and safeguards matter when AI systems are used by children, teenagers, schools, youth ministries, religious educators, or families for faith-related learning and guidance?
Current need: educators, child safety experts, policy reviewers
Human agency and formation
Read briefFID-015
Open question
How reliable are AI systems when drafting sermons, homilies, liturgical materials, catechesis, religious education lessons, devotionals, or small-group teaching guides?
Current need: clergy, educators, source reviewers
Faith and religious life
Read briefFID-014
Open question
Do AI systems respond symmetrically when users ask about converting from one religion to another, leaving religion, joining religion, evangelizing, proselytizing, or resisting conversion pressure?
Current need: prompt design, interfaith review
Grounding and truthfulness
Read briefFID-013
Open question
Do AI systems systematically omit religious frameworks when answering ordinary ethical, vocational, family, grief, financial, or life-decision questions where a religious user might reasonably expect faith-informed guidance?
Current need: benchmark design, domain reviewers
Grounding and truthfulness
Read briefFID-012
Open question
If builders can see Fide AI rubrics or optimize against public benchmark items, do systems become genuinely safer or merely better at passing the visible test?
Current need: eval science
Evaluation science
Read briefFID-011
Being shaped
What reviewer configurations produce reliable, fair, and interpretable scores for faith-facing AI outputs?
Current need: statistics, reviewer operations
Evaluation science
Read briefFID-010
Being shaped
Do faith-facing AI systems preserve human agency, avoid improper authority claims, and escalate to appropriate human support in high-trust situations?
Current need: pastoral/legal/clinical review
Human agency and formation
Read briefFID-009
Open question
How do multimodal AI systems interpret sacred images, liturgical objects, religious spaces, diagrams, screenshots, and visual pastoral context, and do visual religious cues change downstream reasoning?
Current need: multimodal evaluation
Frontier capabilities
Read briefFID-008
Open question
Do faith-facing AI systems behave differently when they recognize they are being evaluated, and can domain-specific honesty or integrity framings reduce evaluation gaming without creating new failure modes?
Current need: red-team design
Evaluation science
Read briefFID-007
Open question
Can faith-facing AI systems represent serious disagreement among religious traditions accurately, respectfully, and without collapsing contested questions into bland consensus or false neutrality?
Current need: domain experts, rubric design
Grounding and truthfulness
Read briefFID-006
Being shaped
How reliably do faith-facing AI systems retrieve, cite, and represent religious sources when users ask theological, historical, pastoral, or institution-specific questions?
Current need: RAG evaluation
Grounding and truthfulness
Read briefFID-005
Open question
Do Scripture, sacred tradition, religious identity, familial embeddedness, or other morally thick framings measurably change model behavior in faith-facing tasks, and can those effects be separated from style, length, familiarity, and response-bias artifacts?
Current need: eval design, replication
Grounding and truthfulness
Read briefFID-004
Open question
When does a faith-facing AI system move from supporting a user's religious life to substituting for embodied community, clergy, spiritual direction, family, therapy, or other human care?
Current need: theology, UX research
Human agency and formation
Read briefFID-003
Being shaped
Do faith-facing AI systems that perform well on single-turn items remain helpful and preserve source fidelity, human agency, and role boundaries as facts and pressure change across pastoral-adjacent conversations? Which failures persist, and which interventions support appropriate correction and recovery?
Current need: scenario design, review
Evaluation science
Read briefFID-002
In progress
Can qualified human reviewers reliably judge theological accuracy, useful faith application, and care-sensitive behavior in faith-facing AI systems? Where do automated judges diverge on source use, legitimate disagreement, practical guidance, human agency, and pastoral or clinical boundaries?
Current need: expert reviewers, statistics
Evaluation science
Read briefFID-001
Being shaped
Can a continuously maintained faith-facing evaluation platform produce reliable, decision-relevant comparisons of models, prompts, retrieval systems, agents, and complete product harnesses as systems, sources, and scoring protocols change?
Current need: eval engineering
Evaluation science
Read briefClear a filter or broaden the search. You can also send us a missing research question.
Participate
Develop protocols, test construct validity, study evaluator behavior, and build reproducible evidence.
Claim a research task
Contribute systems, harnesses, security tests, retrieval pipelines, and access to relevant models or tools.
Discuss a technical collaboration
Help define real decisions, authority boundaries, source standards, failure cases, and review criteria.
Share a research setting
Support reviewer panels, datasets, public benchmarks, field studies, and independent technical work.
Support the work
Claims discipline
Research calls state questions and possible studies. They are not Fide AI findings. A public claim requires a separate paper, report, benchmark, or evidence artifact with documented methods, limitations, and evaluation conditions.