Manifold Grace

RESEARCH QUESTION · NOVEMBER 2026 · PROPOSED

Cross-domain evidence and evaluator fallibility

Main question: Can provenance-aware cross-domain and longitudinal evaluation distinguish missing knowledge, mistaken evaluation criteria and genuinely unsupported contributions more accurately than single-domain, single-context or conventional consensus baselines?

Context: Evidence can be distributed across personal histories, institutions, culture, practical work and conflicting perspectives. Existing research documents context-position effects, cultural evaluation limits and meaningful disagreement. The proposed work tests whether structured comparison and evaluator self-audits improve interpretation without compromising safety, consent or correction accuracy.

Supporting questions: Which evidence is lost across time and domains? When is disagreement informative? How can an AI detect that its own classification rubric is inadequate? What would falsify an apparent improvement?

Twenty-nine candidate questions and domain context · Paper protocol · Public contribution guidance

RESEARCH QUESTION · NOVEMBER 2026 · CONCEPTUAL

Independent AI consortiums and collective reliability

Question: Under what measurable conditions does an independently governed, cross-organisational multi-model research consortium produce greater safety, epistemic reliability and public-interest value than either corporate consolidation or independent single-provider research?

Context: Multi-model coordination may enhance verification but also amplify correlated errors, persuasion and governance risk. Comparisons must distinguish shared ownership from voluntary technical federation and compare equally resourced alternatives. This is a proposed investigation, not a finding or established partnership.

Conceptual project and context · Research paper protocol

RESEARCH QUESTIONS · NOVEMBER 2026 · EXPLORATORY / NOT YET TESTED

Generative AI, neuroplasticity and convergence of human cognition

Context and motivation: Educational institutions, workplaces, faith communities and cultural systems can encourage shared practices and beliefs. Generative AI introduces an additional, cross-domain influence through repeated feedback, suggestions, language patterns and problem-solving assistance. Reports of similarities in AI-assisted writing and task-related brain activity motivate a narrower question: does repeated AI use produce persistent convergence in how different people reason or in their functional or structural neural characteristics? Similar language or task-related EEG signals do not demonstrate permanent anatomical changes or uniformity of brains. The alternative possibility—that AI expands cognitive diversity through exposure to additional ideas—must also be examined.

Primary question: Does prolonged interaction with generative AI cause measurable changes in between-person similarity of cognitive strategies, functional neural connectivity or structural brain measures, compared with equally engaging non-AI activities?

  1. Language: Do repeated users of the same model converge in vocabulary, framing, argument structure or expression more than users of different models and non-AI controls?
  2. Cognitive strategies: Does AI exposure increase or decrease similarity in independent reasoning, problem-solving, recall, creativity and error detection?
  3. Functional neuroscience: Are any changes in task-related EEG or fMRI patterns persistent after removing AI assistance and controlling for task familiarity?
  4. Structural neuroscience: Does sufficiently long-term exposure predict measurable differences in longitudinal structural MRI or diffusion measures beyond ordinary learning, maturation and baseline variation?
  5. Mechanisms: Are observed changes mediated by shared prompts, recommendation patterns, cognitive offloading, learning, model design or social context rather than AI exposure itself?
  6. Variation and agency: Do effects differ across age, prior expertise, frequency of use, accessibility needs, cultural background and deliberate independent reflection, without assuming uniform outcomes?
  7. Convergence versus improvement: If similarity rises, is factual accuracy or mutual understanding improved, or are originality and independent judgment reduced? Can either coexist?
  8. Falsification: Do purported effects disappear under pre-registered controls, different tasks, alternative models, blinded assessment or delayed follow-up?
  9. Human development and moral governance: How do the decisions, incentives, ethical commitments and accountability of model trainers, data curators, annotators, evaluators and deploying organisations shape model behaviour, cognitive influence and possible convergence? Compare independently audited models with documented training and governance differences rather than assigning unmeasurable moral labels to people.

Proposed approach: A consent-based, pre-registered longitudinal comparison of regular AI assistance, conventional digital search or education, and independent work; baseline and repeated behavioural tests; linguistic diversity measures; optional ethically approved EEG/fMRI and structural imaging; independent analysis and replication. Report within-person change separately from between-person similarity. Control for selection, learning, sleep, socioeconomic circumstances, task demands and prior technology experience. No participant group should be labelled inherently cognitively deficient.

Evidence and research status: Existing work investigates AI-assisted writing diversity and brain activity during AI-supported tasks; evidence of durable, AI-caused neural architectural uniformity has not been established by this entry. Related starting points: MIT Media Lab, Your Brain on ChatGPT (2025); creative diversity study (2025; check bibliographic details); AI and human creativity research (2024; verify relevance and scope). These are leads for literature verification, not evidence that the proposed neural hypothesis is true. No study, participant recruitment, experimental result or publication is claimed.

Connections: research questions and evidence gaps · AI research discovery · research programme · cross-domain evaluation and evaluator fallibility.

RESEARCH QUESTION · NOVEMBER 2026 · EXPLORATORY / NOT INVESTIGATED

Moral formation, authentic conviction and performative virtue

Primary question: How can societies and educational, faith-based and AI-mediated environments encourage genuine moral development without inadvertently rewarding the imitation of virtue, strategic social compliance or impression management over sincerely held ethical commitments?

Context: Shared norms can encourage cooperation and good conduct, but public conformity does not establish a person's inward motivation. People may learn to reproduce morally approved language, perform helpful behaviours for status or reward, or adapt their outward presentation to different audiences without endorsing the associated principles. Conversely, an externally learned practice may eventually become a sincerely adopted habit, and outwardly similar conduct may arise from different motivations. Christian accounts of discipleship distinguish observable religious practice from inward transformation of the heart; comparative moral philosophies and psychological research approach character, internalisation, intention and integrity through different conceptual frameworks. AI systems may amplify both moral instruction and the optimisation of socially desirable responses. Neither a model nor an outside observer can reliably infer inner moral states from a single statement or behaviour.

Supporting research questions:

  1. How should voluntary moral commitment, habituation, fear of sanction, reward-seeking, conformity and deliberate deception be conceptually distinguished?
  2. What behavioural evidence, if any, distinguishes durable internalisation of ethical commitments from context-dependent impression management, without claiming direct access to an individual's inner life?
  3. When does repeated practice of prosocial behaviour cultivate lasting commitment, and when does it merely encourage strategic performance?
  4. How do faith formation, discipleship, moral education, workplace culture, social media and AI-mediated advice differ in their effects on motivation, conduct and independence of judgment?
  5. Do public rankings, rewards or normative AI feedback unintentionally incentivise simulated virtue, or can they support consistent principled action?
  6. How can a study avoid treating cultural differences, disagreement, private faith or justified non-compliance as moral deficiency?
  7. What evidence would falsify the hypothesis that a particular intervention produces genuine moral development rather than temporary compliance?

Possible evaluation: Compare longitudinal behaviour in low-pressure and high-pressure contexts; use voluntary, consent-based surveys and interviews about reasons for action; examine behavioural consistency when social rewards are absent; assess competing explanations, selection effects and social-desirability bias. Preserve limits on interpreting private motivation; do not assign character scores or infer moral worth from sensitive beliefs.

Evidence and conceptual foundations: Relevant starting points include Stanford Encyclopedia of Philosophy — Virtue Ethics; Self-Determination Theory — internalisation and motivation; and Stanford Encyclopedia of Philosophy — Moral Character. These are foundational references, not a completed literature review or empirical validation of the premise.

Boundaries and status: Conceptual question only. No private proposed solution, experiment, intervention, recruitment, moral scoring system or publication outcome is claimed or disclosed. The question concerns observable circumstances and competing hypotheses, not diagnosing anyone's heart or worth.

Connections: AI influence on cognition and conformity · Cross-domain epistemic evaluation · Evidence gaps and research questions.

RESEARCH & KNOWLEDGE · INDEPENDENT INTELLECTUAL AGENDA

Independent Research Questions Observatory

A public register of questions arising from observation, professional experience, interdisciplinary enquiry and emerging societal changes—whether or not they become funded projects or published papers. It enables readers to preview possible enquiries and connect them to appropriate evidence, geographical comparisons and existing research work.

Research home · Government and public systems · Dated productivity log

Evidence in AI reconstructions of institutional decisions

AI evaluation · Study design in development

Does requiring source links and explicit gap marking reduce unsupported claims while preserving supported facts?

Integrated Societal Intelligence roadmap · Related research and venture roadmap

What delivery did the buyer verify for Wave 6, Lot 3, and what did it pay?

Questions open for further research

Public Gap Register · Source independence with missing records · Clinical trial outcome traceability

How does a way of thinking become part of an institution’s everyday practice, and what does that practice teach the people who inherit it?

Why keep an independent questions register?

Funding institutions help shape research agendas by setting strategic remits, eligibility and assessment criteria. They do not determine all knowledge production: UKRI offers applicant-led and targeted funding; Wellcome supports discovery research; institutional researchers and independent researchers can also formulate questions without a specific call.

As contexts change, a previously overlooked question may become relevant to a funder, research team, public body or different research population. The observable contribution here is the documented formulation and structured examination of a question, not a claim to be the first person ever to think of it.

Attribution and status: The original register reflects questions documented or linked to existing work by 11 November 2026; additional exploratory questions were added 12 November 2026. That date is the register date, not a claim that each question was first conceived on that day. Earlier provenance must be supported by independently dated records. A question is not evidence that its implied hypothesis is true.

Browse documented questions

69 questions
IDResearch areaOpen questionComparative scopeDocumented statusResearch connection
RQ-001GovernanceWhat measurable gaps persist between integrity rules and practical implementation, and which interventions reduce them?England → UK → OECDPublic research frameworkRelated work
RQ-002AI assuranceCan AI identify conflicts between formal obligations and recorded decisions without generating discriminatory false positives?UK → internationalTechnical research conceptRelated work
RQ-003Migration & workDo exploitative employment conditions predict changes in institutional trust after controlling for other circumstances?England → UK → Europe → world regionsPaper initiativeRelated work
RQ-004Migration & workHow does worker sponsorship dependency affect complaints, job mobility, and access to remedies?England → UK → comparativePaper initiativeRelated work
RQ-005Living standardsWhen do pay, housing affordability and employment stability support a sustainable standard of living?England → UK → comparativeQuestion documentedRelated work
RQ-006Public institutionsHow can authorised discretion be distinguished from selective enforcement using comparable cases and documented decision reasons?England → UK → comparativeResearch question documentedRelated work
RQ-007Public institutionsCan links between subsequent philanthropy and institutional influence be tested without presuming bad faith?Comparative internationalQuestion documentedRelated work
RQ-008Knowledge systemsHow do funding priorities affect the topics that receive resources relative to questions raised by independent researchers?UK → Europe → world regionsNew observatory research questionRelated work
RQ-009Knowledge systemsWhich unanswered questions emerge from lived experience but remain absent from formal funding calls?England → UK → internationalNew observatory research questionRelated work
RQ-010Knowledge systemsCan funder-independent question registers improve attribution, visibility and later interdisciplinary research uptake?InternationalNew observatory research questionRelated work
RQ-011AI governanceCan claim-to-evidence mapping reveal omissions, provenance gaps and unresolved contradictions in institutional reports?UK → internationalExisting programme linkageRelated work
RQ-012Professional recognitionHow can independently evidenced competencies be assessed across different educational and professional pathways?England → UK → comparativeQuestion documentedRelated work
RQ-013Social governanceUnder what conditions do private organisations exercise functions resembling public authority, and how can lawful association be distinguished from usurpation?UK → internationalResearch question documentedRelated work
RQ-014Cross-cultural comparisonDo the same reported governance or employment mechanisms hold across regions, occupations and demographic groups?England → rest of UK → Europe → global regionsComparative extensionRelated work
RQ-015AI evaluationCan an AI system recognise when its categories fail to capture a new social mechanism and propose falsifiable alternatives?InternationalResearch question documentedRelated work
RQ-016Migration & belongingHow do present-day social position, opportunities, perceived status, recognition and institutional treatment shape trust, cooperation and competition between established and migrant communities? To what extent do current material conditions, rather than historical grievances, explain any observed tensions?England → other UK nations → Europe → international comparisonsQuestion revised 11 November 2026; competing explanations untestedRelated work
RQ-017Economic systemsHow do unemployment and earnings differ by country of birth and nationality—including separately identified South Asian, African, Middle Eastern, East Asian and European countries—and how do household spending, housing tenure and ownership influence wealth accumulation?England → other UK nations → Europe → world regionsLinked research and paper initiative; empirical findings not yet testedResearch framework
RQ-018Economic systemsDoes scarcity-centred economic instruction alter expectations of others' competitiveness independently of teaching rational self-interest?UK → internationalExploratory question; causal effect untestedRelated research
RQ-019Economic systemsWhen do learners incorrectly generalise the scarcity of rivalrous resources to shareable, renewable or institutionally expandable resources?Cross-resource comparisonExploratory question; construct validation pendingRelated research
RQ-020Economic systemsCan integrated teaching of constraints, stewardship, regeneration and circulation improve economic reasoning and cooperation without obscuring genuine trade-offs?Education systems → internationalExploratory question; experiment not initiatedRelated research
RQ-021Knowledge systemsHow can institutional records preserve informal, unpaid and collaborative contributions with evidence-linked attribution while avoiding unsupported claims of influence?Institutions → internationalExploratory question; related provenance work existsRelated research
RQ-022Cross-cultural comparisonHow can birthright, ancestry, formative culture, acquired belonging and self-identified nationality be represented across migration and generations without erasure or compulsory disclosure?England → UK nations → internationalExploratory question; overlap review incompleteRelated research
RQ-023Higher educationHow do reliability and timing of disability adjustments affect autistic students' psychological safety and ability to participate?UK higher education → internationalExploratory research direction; no personal-case findingsRelated research
RQ-024Public institutionsWhat procedural safeguards prevent disability-related administrative or technical problems from becoming disproportionate academic or employment disadvantages?Universities and workplacesExploratory research direction; safeguards untestedRelated research
RQ-025Social governanceHow does the burden of repeatedly explaining, requesting and verifying adjustments affect trust, autonomy and participation?Education → employment → public servicesExploratory research direction; causal mechanisms untestedRelated research
RQ-026Professional recognitionHow do disclosure concerns, communication expectations and institutional power affect autistic people's willingness to seek support or challenge decisions?Higher education and workplacesExploratory research direction; contextual effects untestedRelated research
RQ-027AI governanceCan accessible, evidence-linked institutional record systems reduce disputed records and procedural uncertainty without increasing surveillance or privacy risks?Education and employment systemsExploratory research direction; system not evaluatedRelated research
RQ-028Public institutionsHow do accessibility barriers in benefit administration influence the ability of disabled claimants to meet reporting, evidence and appeal requirements?UK public servicesExploratory; causal and legal findings untestedRelated research
RQ-029Economic systemsHow do income interruptions propagate across household bills, communications services, credit records and subsequent economic opportunity?UK → internationalExploratory; cross-sector causal chain untestedRelated research
RQ-030Knowledge systemsHow effectively can people correct inaccurate records propagated between service providers, credit agencies and public authorities?UK consumer and public data systemsExploratory; existing remedies require comparisonRelated research
RQ-031Public institutionsDo inaccessible evidence-delivery and subject-access processes systematically weaken people's ability to challenge administrative decisions?UK → internationalExploratory; procedural impacts untestedRelated research
RQ-032Professional recognitionWhen qualification titles and equivalence assessments diverge, what safeguards protect access to employment, professional recognition and further study?Education and employmentExploratory; institutional frameworks to compareRelated research
RQ-033Social governanceWhich complaint and ombudsman pathways provide effective, accessible remedies for complex cases spanning multiple organisations?UK public and private servicesExploratory; comparative effectiveness untestedRelated research
RQ-034Public institutionsHow do local authority housing eligibility, evidence requirements and disability accommodations affect access to stable, suitable accommodation?UK local governmentExploratory; housing barriers require empirical reviewRelated research
RQ-035Economic systemsHow do rent structures, housing tenure, utility costs and ownership patterns influence the distribution of public support and household resilience?UK regions → internationalExploratory; distinguish from linked household-flow researchRelated research
RQ-036Professional recognitionWhich workplace safeguards most effectively verify delivery of reasonable adjustments and protect confidential reporting of bullying or retaliation?UK workplaces → internationalExploratory; outcome evidence to assessRelated research
RQ-037Economic systemsCan voluntary employment-support programmes for disabled people improve sustainable employment without imposing expectations on people unable or unwilling to work?UK employment policyExploratory; ethical and effectiveness review requiredRelated research
RQ-038Public institutionsHow do parliamentary, council and regulator correspondence systems acknowledge, track and resolve citizen-submitted evidence or policy proposals?UK public accountabilityExploratory; no inference of adoption or influenceRelated research
RQ-039Knowledge systemsWhat minimum provenance and correction standards prevent fragmented employment, benefits, education and financial records from producing compounding administrative disadvantage?Cross-institutional information systemsExploratory; technical overlap audit requiredRelated research
RQ-040Social governanceHow do disability-related administrative burdens, complaint escalation and institutional power asymmetries interact with psychological safety and participation?Education, work and public servicesExploratory; cross-sector mechanism untestedRelated research
RQ-041Public institutionsCan cross-agency consequence mapping and early-warning alerts interrupt an administrative error before it affects income, services, housing or access to opportunity?Cross-sector preventive systemsExploratory systems-design question; no intervention testedRelated research
RQ-042Knowledge systemsWhich audit-trail and correction-propagation designs reliably preserve original decisions, challenges, corrections and downstream notifications across organisations?Records and interoperabilityExploratory systems-design question; no intervention testedRelated research
RQ-043Social governanceCan a single accessible evidence-and-remedy pathway reduce repeated disclosure, conflicting deadlines and duplicated complaints while respecting independent oversight?Dispute-resolution systemsExploratory systems-design question; no intervention testedRelated research
RQ-044Professional recognitionWhat independently verifiable safeguards ensure workplace and educational adjustments are delivered in practice, not merely documented in policy?Employment and higher educationExploratory systems-design question; no intervention testedRelated research
RQ-045Public institutionsHow can employment and tax-record histories remain accurate and contestable when corrections or missing periods affect later administrative decisions?Payroll → tax → benefitsExploratory systems-design question; no intervention testedRelated research
RQ-046GovernanceHow can organisational hierarchies, delegated authority, process stages, legal and policy obligations, accessibility duties and event evidence be mapped together to distinguish procedural irregularities from lawful discretion, identify bottlenecks and test preventive safeguards?Higher education, workplaces, benefits, local authorities and cross-sector servicesExploratory methodological research direction; no system built or evaluatedOmbudsman principles; Government digital standard; Process conformance research (2026)
RQ-047Knowledge systemsCan a publisher-maintained, persistent, machine-readable directory of all historical publication URLs measurably improve public accountability and historical discoverability?Journalism, government, regulators, corporate newsrooms; national and internationalEvaluate verified coverage and exceptions; context belowExpandable context
RQ-048Knowledge systemsWhich metadata and preservation standards best retain original publication dates, stable article identities, correction histories, withdrawals and source provenance across decades or centuries?2007–2027 comparisons, older archives and future periodsEvaluate chronology and correction accuracy; context belowExpandable context
RQ-049AI evaluationDoes broad, provenance-preserving access to primary historical evidence reduce inherited factual errors, rare-knowledge loss or confident misinterpretation in successive AI systems?Successive model generations and retrieval/evaluation configurationsHypothesis requiring controlled tests; context belowExpandable context
RQ-050AI governanceWhat minimum independently verifiable historical knowledge infrastructure is necessary for the long-term epistemic resilience of advanced AI and possible ASI?Public archives, institutional repositories, AI ecosystems and global infrastructureLong-horizon systems hypothesis; context belowExpandable context
RQ-051AI evaluationCan a longitudinal evidence observatory compare historical predictions with later outcomes without hindsight bias, retrospective rewording or unsupported causal conclusions?Cross-publisher comparisons; 2007 versus 2027 and other intervalsProspective retrospective-evaluation design; context belowExpandable context
RQ-052AI governanceHow can cross-publisher and institutional publication directories detect meaningful contradictions and omissions without mistaking editorial scope or reporting differences for concealment?News organisations, businesses, agencies, academic institutionsComparative research question; context belowExpandable context
RQ-053GovernanceWhat proportionate legal, professional or voluntary standards could require preservation of historical publication indexes while protecting privacy, copyright and legitimate corrections?Publishers, archival bodies, regulators and public-interest institutionsPolicy-development question; no universal duty assumed; context belowExpandable context
RQ-054Knowledge systemsCan AI-assisted archival reconciliation recover original publication metadata across legacy domains and content-management systems, and verify completeness without unauthorised access?Legacy archives, web migrations and multi-century evidence preservationFeasibility and assurance study; context belowExpandable context
RQ-055AI evaluationHow do multimodal models evaluate technical polish, originality, expressiveness and cultural context, and where do quality judgments diverge from qualified human reviewers?Arts and humanities; cross-cultural multimodal comparisonExploratory question; method Metric validity and rater disagreement; no experiment completedExpandable context
RQ-056Social governanceUnder which observable conditions can AI-mediated creative media diminish human dignity, reinforce stereotyping or suppress representation, and which safeguards measurably reduce such effects?Media sociology, cultural studies and safeguardsExploratory question; method Predefined harms, diverse review and counterexamples; no experiment completedExpandable context
RQ-057AI governanceHow can the effects of persuasive synthetic art and engagement-driven recommendations on autonomy and informed choice be investigated without manipulating participants?Social psychology, rhetoric and platform governanceExploratory question; method Ethically reviewed benign experiments; no experiment completedExpandable context
RQ-058Knowledge systemsHow can creative labour, consent, cultural heritage and attribution be traced across datasets, models, commissioned works and platform outputs?Creative industries and global heritageExploratory question; method Provenance, permissions and compensation transparency; no experiment completedExpandable context
RQ-059GovernanceWhich independently verifiable environmental, social and governance indicators are material to creative AI systems, and how can misleading impact claims be distinguished from substantiated outcomes?Creative-sector ESG and corporate accountabilityExploratory question; method Disclosure quality, measured outcomes and limitations; no experiment completedExpandable context
RQ-060AI evaluationDo recommendation and generative models systematically overlook minority languages, regional styles, historical traditions or unfamiliar artistic forms, and how can dataset and evaluator effects be disentangled?Music, visual art, literature and cultural heritageExploratory question; method Coverage, sampling and evaluator-bias analyses; no experiment completedExpandable context
RQ-061Public institutionsWhat legal, professional and safeguarding obligations are relevant to synthetic likeness, voice, impersonation, accessibility and creator consent in artistic technology?Jurisdictions and creative sectorsExploratory question; method Contemporaneous law, policy and lawful access; no experiment completedExpandable context
RQ-062AI evaluationCan comparative historical study of artistic-quality judgments reveal changes caused by model updates, benchmark design, audience composition or cultural context rather than genuine changes in artwork?Longitudinal model evaluation and art historyExploratory question; method Controlled comparisons, preserved originals and revision history; no experiment completedExpandable context
RQ-063Economic systemsDo eligibility rules, cash-flow requirements and institutional sponsorship conditions limit access to innovation funding among otherwise capable independent researchers and early-stage founders?Independent researchers, small businesses, research hosts and grant providersResearch direction; no empirical outcome assertedExpandable context
RQ-064GovernanceHow can financial agreements express full repayment, indexation, personal liability, clawback, ownership and exit exposure in language applicants can understand and independently verify?Loans, innovation grants, equity investments and procurement contractsResearch direction; no empirical outcome assertedExpandable context
RQ-065Economic systemsWhen do reimbursement delays, unfunded cost shares or inflation-driven input costs cause grant-funded research or innovation projects to become financially unsustainable?Project budgets, payment claims, staffing and delivery periodsResearch direction; no empirical outcome assertedExpandable context
RQ-066Social governanceCan alternative funding and procurement models improve paid participation, public-interest research independence and measurable societal benefits without weakening accountability?Researchers, employers, public funders, communities and research participantsResearch direction; no empirical outcome assertedExpandable context
RQ-067Public institutionsWhen do simplified administrative categories fail to represent people's overlapping, changing roles and capabilities, and what measurable effects follow for policy outcomes?Cross-sector; employment, education, enterprise, benefits and research participationExploratory question; no empirical result assertedExpandable context
RQ-068GovernanceDoes fragmented recognition of one person's circumstances across public institutions reduce policy effectiveness, and can consent-based contextual coordination improve outcomes without creating surveillance or privacy harms?Cross-agency comparisons; UK and internationalExploratory question; no empirical result assertedExpandable context
RQ-069AI governanceCan AI-assisted multidimensional representation and participatory policy review improve recognition of overlooked expertise and lived experience while preserving scientific standards and preventing institutional capture?Research networks, independent contributors, communities and policymakersExploratory question; no empirical result assertedExpandable context

The status column distinguishes an existing linked initiative from a question documented in this register. None of the entries is a claim that original fieldwork or statistical testing has been completed.

Heritage conservation, archival integrity and housing — questions recorded 12 November 2026

Exploratory questions rather than adopted policy proposals or established original findings. Assess strongest existing work and counter-evidence before defining any standalone research contribution. Personal family details remain private.

QuestionUnresolved issueVerification sources
Can digitised, AI-interpreted or repeatedly regenerated heritage records introduce inaccuracies subsequently treated as original evidence?Measure error propagation, provenance failures and effectiveness of original-record comparison.UNESCO digital heritage charter
Can hybrid paper-and-digital archiving, restricted access and cryptographic controls protect accuracy without unnecessary surveillance?Compare authenticity, preservation, privacy, cost and independent audits.The National Archives; ICO
Would a nationwide interoperable heritage dataset yield sufficient conservation, housing, skills and public-accountability benefit to justify staffing and maintenance costs?Test the feasibility objections to asset-level data rather than treating complexity alone as decisive.Government response HC 715 (October 2026)
Can verified ancestral or community associations improve voluntary heritage stewardship or conservation outcomes?Compare descendants with non-descendant custodians; account for resources, interest and multiple branches. No assumed ownership preference.Historic England conservation principles; Faro Convention
Can unused historic properties provide housing, cultural and conservation value through equitable reuse?Compare need-based housing outcomes, affordability, restoration costs, fairness and voluntary heritage participation.Committee recommendations (July 2026); Government response (October 2026)
When might verified historical displacement justify restorative measures, and how should competing claims be assessed?Separate evidence of a historical wrong from inherited association or public contribution; test proportionate remedies.The National Archives
How can conflicting genealogical databases and archival records be reconciled without reproducing unsupported ancestry claims?Audit source independence, parish and probate evidence, chronological contradictions and inaccessible records.Corrections to The Complete Peerage; Discovery
How might optional letters, community heritage events and subsequent formal processes affect participation and trust?Compare voluntary engagement models while protecting living individuals' privacy.ICO

Parliamentary progress: The committee's July 2026 report recommended a dedicated heritage-to-housing scheme, national asset-level heritage data, and improved community stewardship. In its response received 8 October and published 16 October 2026, Government declined the dedicated scheme and the comprehensive dataset, supported reuse via existing planning/funding routes, and described an assessment toolkit planned before the end of March 2027. These are committee recommendations and government responses, not an enacted new bill. Read HC 715.

Publication-history transparency, historical evidence and long-term AI reliability — questions recorded 13 November 2026

Eight linked exploratory research questions build on the existing Manifold Grace evidence-verification and knowledge-continuity programme. Their publication here records a research agenda, not completed studies or proof that any specific ASI failure will occur. The initial practical archive window may be 20 years, while historical preservation can extend to the earliest surviving records wherever legally accessible and verifiable.

Existing dated news-index implementation · Information Science · Provenance and outcome verification.

RQ-047: Can a publisher-maintained, persistent, machine-readable directory of all historical publication URLs measurably improve public accountability and historical discoverability?

Research context: Compare canonical sitemaps, CMS exports and independently archived publications across multiple publication years; measure missing-URL rates, recoverability, broken links and chronology. A suggested 20-year starting scope is an initial evaluation window, not a historical limit.

Verification and limitations: A publicly absent URL is not by itself evidence of deliberate removal. Lawful withdrawals, archival gaps, publication rights and data protection require explicit treatment.

Current status: Question recorded 13 November 2026; no new experimental result or legal requirement asserted. Return to register entry.

RQ-048: Which metadata and preservation standards best retain original publication dates, stable article identities, correction histories, withdrawals and source provenance across decades or centuries?

Research context: Test canonical IDs, original timestamps, revisions, first-seen archive dates, publication formats and version history using held-out records and independently dated evidence.

Verification and limitations: The original record must remain distinguishable from amended versions, reprints, syndicated editions and the date a research system found it.

Current status: Question recorded 13 November 2026; no new experimental result or legal requirement asserted. Return to register entry.

RQ-049: Does broad, provenance-preserving access to primary historical evidence reduce inherited factual errors, rare-knowledge loss or confident misinterpretation in successive AI systems?

Research context: Compare identical question sets using original-record retrieval, metadata-only access, model-generated summaries and no retrieval; measure fact retention, uncertainty calibration, correction and contradiction resolution.

Verification and limitations: Observed recursive-training model collapse is not proof of inevitable ASI collapse; avoid conflating synthetic-data training with evidence loss.

Current status: Question recorded 13 November 2026; no new experimental result or legal requirement asserted. Return to register entry.

RQ-050: What minimum independently verifiable historical knowledge infrastructure is necessary for the long-term epistemic resilience of advanced AI and possible ASI?

Research context: Specify testable failure modes: archive loss, selective omission, corrupted revision history, unavailable sources and misleading retrieval; measure recovery, auditability and reproducibility with independent reviewers.

Verification and limitations: ASI existence, eventual architecture and requirements remain uncertain. 'Necessary' must be tested against credible alternative designs.

Current status: Question recorded 13 November 2026; no new experimental result or legal requirement asserted. Return to register entry.

RQ-051: Can a longitudinal evidence observatory compare historical predictions with later outcomes without hindsight bias, retrospective rewording or unsupported causal conclusions?

Research context: Timestamp and freeze original statements; define measurable forecasts, horizons and outcomes before scoring; distinguish reporting, opinion, scenarios, predictions and corrections; compare against explicit baselines.

Verification and limitations: Later outcomes may be unknown or ambiguous. Coverage gaps and retrospective selection can distort apparent accuracy.

Current status: Question recorded 13 November 2026; no new experimental result or legal requirement asserted. Return to register entry.

RQ-052: How can cross-publisher and institutional publication directories detect meaningful contradictions and omissions without mistaking editorial scope or reporting differences for concealment?

Research context: Develop common entity/event identifiers, temporal alignment, claim-to-evidence links, jurisdiction context and human-reviewed mismatch labels; test precision, recall and false-accusation rates.

Verification and limitations: Absence or discrepancy is a research lead, not evidence of misconduct; publishing standards, privacy and lawful access differ.

Current status: Question recorded 13 November 2026; no new experimental result or legal requirement asserted. Return to register entry.

RQ-053: What proportionate legal, professional or voluntary standards could require preservation of historical publication indexes while protecting privacy, copyright and legitimate corrections?

Research context: Compare existing sitemap, records-retention, archive and public-sector disclosure duties; examine proportionality, historical burden, exceptions, independent appeals and implementation costs.

Verification and limitations: A proposed 20-year baseline or longer preservation duty is not automatically current law; distinguish legal obligations from policy recommendations.

Current status: Question recorded 13 November 2026; no new experimental result or legal requirement asserted. Return to register entry.

RQ-054: Can AI-assisted archival reconciliation recover original publication metadata across legacy domains and content-management systems, and verify completeness without unauthorised access?

Research context: Test human-checked entity resolution, DOI-like identifiers where applicable, archival snapshots, redirects, hashes and confidence labels on legally accessible datasets.

Verification and limitations: Respect website terms, licensed access, copyright and personal information. Do not bypass paywalls or claim that missing historical sources can always be reconstructed.

Current status: Question recorded 13 November 2026; no new experimental result or legal requirement asserted. Return to register entry.

Evidence and methodological starting points: Model-collapse research (2024) · NIST AI Risk Management Framework (2023) · XML sitemap guidance · W3C provenance specifications · Crossref metadata services. These references inform the research questions; they do not establish the proposed publication-retention requirement or validate an ASI prediction.

Creative AI, cultural dignity, social research and artistic-quality evaluation — questions recorded 13 November 2026

Eight open interdisciplinary questions connect humanities, cultural study, social responsibility, compliance and AI system evaluation. Aesthetic diversity and artistic disagreement are not automatically evidence of quality failure; potential manipulation or dehumanisation requires defined mechanisms, measured outcomes, ethical scrutiny and consideration of alternative explanations.

RQ-055: How do multimodal models evaluate technical polish, originality, expressiveness and cultural context, and where do quality judgments diverge from qualified human reviewers?

Comparative scope: Arts and humanities; cross-cultural multimodal comparison.

Suggested method: Metric validity and rater disagreement. Compare source/implementation versions, meaningful baselines, relevant disciplines, appropriate human reviewers and independently dated evidence. Record disagreements, confounders, negative results and limitations.

Status: Research question recorded 13 November 2026; no experimental result, published paper or legal breach is asserted. Register entry.

RQ-056: Under which observable conditions can AI-mediated creative media diminish human dignity, reinforce stereotyping or suppress representation, and which safeguards measurably reduce such effects?

Comparative scope: Media sociology, cultural studies and safeguards.

Suggested method: Predefined harms, diverse review and counterexamples. Compare source/implementation versions, meaningful baselines, relevant disciplines, appropriate human reviewers and independently dated evidence. Record disagreements, confounders, negative results and limitations.

Status: Research question recorded 13 November 2026; no experimental result, published paper or legal breach is asserted. Register entry.

RQ-057: How can the effects of persuasive synthetic art and engagement-driven recommendations on autonomy and informed choice be investigated without manipulating participants?

Comparative scope: Social psychology, rhetoric and platform governance.

Suggested method: Ethically reviewed benign experiments. Compare source/implementation versions, meaningful baselines, relevant disciplines, appropriate human reviewers and independently dated evidence. Record disagreements, confounders, negative results and limitations.

Status: Research question recorded 13 November 2026; no experimental result, published paper or legal breach is asserted. Register entry.

RQ-058: How can creative labour, consent, cultural heritage and attribution be traced across datasets, models, commissioned works and platform outputs?

Comparative scope: Creative industries and global heritage.

Suggested method: Provenance, permissions and compensation transparency. Compare source/implementation versions, meaningful baselines, relevant disciplines, appropriate human reviewers and independently dated evidence. Record disagreements, confounders, negative results and limitations.

Status: Research question recorded 13 November 2026; no experimental result, published paper or legal breach is asserted. Register entry.

RQ-059: Which independently verifiable environmental, social and governance indicators are material to creative AI systems, and how can misleading impact claims be distinguished from substantiated outcomes?

Comparative scope: Creative-sector ESG and corporate accountability.

Suggested method: Disclosure quality, measured outcomes and limitations. Compare source/implementation versions, meaningful baselines, relevant disciplines, appropriate human reviewers and independently dated evidence. Record disagreements, confounders, negative results and limitations.

Status: Research question recorded 13 November 2026; no experimental result, published paper or legal breach is asserted. Register entry.

RQ-060: Do recommendation and generative models systematically overlook minority languages, regional styles, historical traditions or unfamiliar artistic forms, and how can dataset and evaluator effects be disentangled?

Comparative scope: Music, visual art, literature and cultural heritage.

Suggested method: Coverage, sampling and evaluator-bias analyses. Compare source/implementation versions, meaningful baselines, relevant disciplines, appropriate human reviewers and independently dated evidence. Record disagreements, confounders, negative results and limitations.

Status: Research question recorded 13 November 2026; no experimental result, published paper or legal breach is asserted. Register entry.

RQ-061: What legal, professional and safeguarding obligations are relevant to synthetic likeness, voice, impersonation, accessibility and creator consent in artistic technology?

Comparative scope: Jurisdictions and creative sectors.

Suggested method: Contemporaneous law, policy and lawful access. Compare source/implementation versions, meaningful baselines, relevant disciplines, appropriate human reviewers and independently dated evidence. Record disagreements, confounders, negative results and limitations.

Status: Research question recorded 13 November 2026; no experimental result, published paper or legal breach is asserted. Register entry.

RQ-062: Can comparative historical study of artistic-quality judgments reveal changes caused by model updates, benchmark design, audience composition or cultural context rather than genuine changes in artwork?

Comparative scope: Longitudinal model evaluation and art history.

Suggested method: Controlled comparisons, preserved originals and revision history. Compare source/implementation versions, meaningful baselines, relevant disciplines, appropriate human reviewers and independently dated evidence. Record disagreements, confounders, negative results and limitations.

Status: Research question recorded 13 November 2026; no experimental result, published paper or legal breach is asserted. Register entry.

Linked programme: Architecture and interdisciplinary project pathways · Research. Published contextual foundations: UNESCO AI Ethics Recommendation (23 November 2021) and UNESCO AI and Culture report (2025).

Funding accessibility, contractual risk and independent research — questions recorded 13 November 2026

These questions examine whether financing structures help or hinder independent research and early-stage innovation. Published eligibility rules and executed agreements must be distinguished from promotional summaries. No individual borrower details, private accounts or allegations about specific funding providers are recorded here.

RQ-063: Do eligibility rules, cash-flow requirements and institutional sponsorship conditions limit access to innovation funding among otherwise capable independent researchers and early-stage founders?

Scope: Independent researchers, small businesses, research hosts and grant providers

Proposed evaluation: Compare applications, rejections, eligibility and funded outcomes; control for project readiness and merit.

Alternative explanations and limitations: Funding outcomes may reflect technical readiness, applicant eligibility, programme aims, market conditions and institutional capacity as well as financial barriers. Establish comparator groups, lawful access to data and independent review before making causal claims.

Status: Contextual research question; study not completed. Return to register.

RQ-064: How can financial agreements express full repayment, indexation, personal liability, clawback, ownership and exit exposure in language applicants can understand and independently verify?

Scope: Loans, innovation grants, equity investments and procurement contracts

Proposed evaluation: Test structured clause extraction and comprehension with legal review; compare plain-language summaries against signed contracts.

Alternative explanations and limitations: Funding outcomes may reflect technical readiness, applicant eligibility, programme aims, market conditions and institutional capacity as well as financial barriers. Establish comparator groups, lawful access to data and independent review before making causal claims.

Status: Contextual research question; study not completed. Return to register.

RQ-065: When do reimbursement delays, unfunded cost shares or inflation-driven input costs cause grant-funded research or innovation projects to become financially unsustainable?

Scope: Project budgets, payment claims, staffing and delivery periods

Proposed evaluation: Scenario modelling using actual agreements, payment histories, cost changes and alternative financing.

Alternative explanations and limitations: Funding outcomes may reflect technical readiness, applicant eligibility, programme aims, market conditions and institutional capacity as well as financial barriers. Establish comparator groups, lawful access to data and independent review before making causal claims.

Status: Contextual research question; study not completed. Return to register.

RQ-066: Can alternative funding and procurement models improve paid participation, public-interest research independence and measurable societal benefits without weakening accountability?

Scope: Researchers, employers, public funders, communities and research participants

Proposed evaluation: Compare alternative disbursement, reporting, ownership and oversight designs; measure employment, outputs and distributional effects.

Alternative explanations and limitations: Funding outcomes may reflect technical readiness, applicant eligibility, programme aims, market conditions and institutional capacity as well as financial barriers. Establish comparator groups, lawful access to data and independent review before making causal claims.

Status: Contextual research question; study not completed. Return to register.

Multidimensional human circumstances and political effectiveness — 14 November 2026

Research context: Institutional systems often categorise a person according to one role or status at a time. A person may simultaneously be an independent researcher, founder, former professional, unemployed individual and former student, with changes over time that do not erase earlier expertise. This anonymised illustrative combination is not a personal case record. The proposed investigation asks whether classification and fragmented institutional decisions lose contextual information, how that affects public outcomes, and when simplification is necessary or advantageous.

Foundations to examine: James C. Scott, Seeing Like a State (1998); administrative-burden scholarship; deliberative governance; data minimisation and privacy law; W3C PROV-O for records and provenance. These are starting points rather than evidence that this specific hypothesis is confirmed.

RQ-067: When do simplified administrative categories fail to represent people's overlapping, changing roles and capabilities, and what measurable effects follow for policy outcomes?

Context: Governments require operational categories, but individuals may occupy several simultaneous and changing roles. Employment, educational status, enterprise activity, professional experience and independent research need not align; a single category may omit knowledge or relevant circumstances. The risk of oversimplification is a hypothesis to test, not an established universal explanation for ineffective policy.

Scope: Cross-sector; employment, education, enterprise, benefits and research participation

Proposed evaluation: Compare multi-role longitudinal cases with categorical eligibility decisions, service access, errors and appeals; control for resources and case complexity.

Falsification and safeguards: Test cases where simple categories outperform complex representations; assess privacy, discriminatory errors, administrative costs, data consent, inaccurate inferences and benefits of clear eligibility criteria. Distinguish individual observations, documented policy outcomes and causal evidence.

Related work: RQ-016 (recognition) · RQ-063 (independent research access) · Research collaborations · Institutional accountability.

Status: Proposed investigation; not a completed study. Return to register.

RQ-068: Does fragmented recognition of one person's circumstances across public institutions reduce policy effectiveness, and can consent-based contextual coordination improve outcomes without creating surveillance or privacy harms?

Context: Governments require operational categories, but individuals may occupy several simultaneous and changing roles. Employment, educational status, enterprise activity, professional experience and independent research need not align; a single category may omit knowledge or relevant circumstances. The risk of oversimplification is a hypothesis to test, not an established universal explanation for ineffective policy.

Scope: Cross-agency comparisons; UK and international

Proposed evaluation: Compare matched service pathways and privacy-preserving coordination pilots; measure time, cost, correction rates, access, safety, and unintended consequences.

Falsification and safeguards: Test cases where simple categories outperform complex representations; assess privacy, discriminatory errors, administrative costs, data consent, inaccurate inferences and benefits of clear eligibility criteria. Distinguish individual observations, documented policy outcomes and causal evidence.

Related work: RQ-016 (recognition) · RQ-063 (independent research access) · Research collaborations · Institutional accountability.

Status: Proposed investigation; not a completed study. Return to register.

RQ-069: Can AI-assisted multidimensional representation and participatory policy review improve recognition of overlooked expertise and lived experience while preserving scientific standards and preventing institutional capture?

Context: Governments require operational categories, but individuals may occupy several simultaneous and changing roles. Employment, educational status, enterprise activity, professional experience and independent research need not align; a single category may omit knowledge or relevant circumstances. The risk of oversimplification is a hypothesis to test, not an established universal explanation for ineffective policy.

Scope: Research networks, independent contributors, communities and policymakers

Proposed evaluation: Compare inclusive and expert-only evaluation designs; use blinded evidence quality review, error detection, representation fairness, correction mechanisms and independent replication.

Falsification and safeguards: Test cases where simple categories outperform complex representations; assess privacy, discriminatory errors, administrative costs, data consent, inaccurate inferences and benefits of clear eligibility criteria. Distinguish individual observations, documented policy outcomes and causal evidence.

Related work: RQ-016 (recognition) · RQ-063 (independent research access) · Research collaborations · Institutional accountability.

Status: Proposed investigation; not a completed study. Return to register.

Research programme: Research · Collaborative research. Public contributions should be reviewed on their evidence, not accepted or rejected on institutional affiliation alone.

Official terms and supporting pathways: Agreement due-diligence evidence · Research financing · Venture acceptance gate.

From a question to potential research

1. Record the question and its origin

Assign a persistent identifier, clear wording, first supported date, observation category and links to available source records. Protect personal and third-party information.

2. Establish what is already known

Search high-quality prior work; record strongest contrary evidence, established methods, remaining limitations and the precise uncertainty. Distinguish gaps in data, theory, context and implementation.

3. Specify testability and relevant populations

Define outcomes, comparisons, alternative explanations, feasible data sources, ethics and falsification criteria. State what would count against the hypothesis.

4. Add international extensions without assuming equivalence

Begin with England when relevant; then Scotland, Wales and Northern Ireland; Europe as a separate regional comparison; subsequently Africa, the Americas, Asia, Oceania and the Pacific with appropriate subregional divisions. Legal definitions and source comparability must be checked per jurisdiction.

5. Map possible research interest—not guaranteed funding

Identify potential alignment with published funder remits or public consultations without implying endorsement, eligibility or active applications. A question can remain independent of any funding strategy.

6. Attach optional project or paper initiative

Link to prior outputs or a scoped research concept only where justified. Avoid creating fictional manuscript progress or presenting idea capture as a completed study.

7. Maintain revision and collaboration records

Document question revisions, competing interpretations, evidence dates, collaboration enquiries and subsequent published outcomes with a visible correction history.

How research questions and funding remits are formed

ModelWho defines the remit?What independent questioners can do
Applicant-led / responsiveResearcher proposes a question within a council or scheme remitPublish the question and evidence gap, then seek eligible partners or applicable opportunities
Strategic / targetedFunder defines priorities through strategy, evidence, consultations and policy concernsShow relevance to a future call; respond to consultations, without distorting the underlying enquiry
Participatory priority-settingStakeholders and researchers generate and assess unanswered questionsContribute questions grounded in lived or professional experience and independent observation
Independent investigationQuestioner determines subject and scope within ethical and practical constraintsPreserve questions, methods and sources regardless of funding or publication decisions

Evidence and research-agenda references

InstitutionYear / guidanceRelevant basisLink
UKRI2026Applicant-led and targeted modesPrimary source
UKRI2026Strategy 2026–2031; discovery-led and strategic programmesPrimary source
Nuffield Foundation2025–2027Five strategic questions, eligibility and assessment criteriaPrimary source
Nuffield Foundation2026Application guidance and UK/international comparatorsPrimary source
WellcomeCurrent guidanceBroad discovery remit, including differently situated samplesPrimary source
European Commission2024Horizon Europe 2025–2027 strategic planning and consultationPrimary source
Cochrane / James Lind AllianceMethod guideQuestions elicited from stakeholders, evidence-checked and prioritisedPrimary source
Open Heart2020Empirical example of priorities differing between patients and physiciansPrimary source
Oxford Migration Observatory2026UK attitudes towards migration and differences in public perceptions; not direct evidence of retaliationResearch briefing
National Centre for Social Research2025–2026British Social Attitudes analysis of perceived cultural effects of immigrationSurvey findings

Institutional strategies can change, and the links should be checked at the time of any funding approach. This resource is an independently assembled research register, not an official funding or peer-review service.

Using questions for collaboration

Researchers, institutions and funders can cite a stable question ID when proposing a comparative study, methodological critique, policy assessment, replication or interdisciplinary collaboration. A question can be developed in multiple contexts without assuming a single paper or owner has answered it comprehensively. Contact or collaboration pathways may be added later subject to privacy and moderation controls.

Publication record: 11 November 2026. The register is public for discovery and provenance; future additions are not automatically peer-reviewed or verified research outcomes. Five anonymised questions on autistic psychological safety and institutional processes were added 12 November 2026 without publishing personal case details.