Manifold Grace

Research & Knowledge · Public Interest AI · Enterprise

Societal Resilience & the AI Public Adviser

An interdisciplinary programme connecting geopolitical developments, the ethics of artificial advisers, probability and prediction, state power, accountable AI, economic adaptation and practical public resilience. It draws upon themes studied at master's level in AI, ethics and society, alongside continuing independent investigation. This page is not a claim that a university module was completed or assessed.

Public adviser status (10 November 2026): A working, browser-only source-comparison demonstration is available below. It is not a deployed language model, live news feed, commercial investment adviser or verified forecast service. It does not scrape, upload or store visitor text on a server.

Research programme: breadth before depth

Ethics of political AI advisers

Human authority, accountable recommendations, institutional power, democratic scrutiny and conflict-sensitive information.

Probability, models and prediction

Calibration, baselines, scenario dependence, missing data, uncertainty and the difference between conditional forecasts and certainty.

The value and materiality of data

Data provenance, unequal access, licensing, physical compute costs, infrastructure and environmental consequences.

Fairness and explainability

Who is represented, who is absent, which assumptions drive outputs and how decisions can be challenged or corrected.

Geopolitical and systemic risk

Conflicts, trade disruptions, critical services, public institutions, supply chains, resilience and humanitarian consequences.

Work, enterprise and public benefit

Labour transitions, cooperative ventures, viable businesses, public goods and meaningful activity outside paid employment.

Cross-link to the history and foundations of AI

Related historical perspectives include symbolic systems, statistical modelling, knowledge representation, machine learning, geopolitical event datasets, and theories of delegation and political authority. View the 22-platform history of societal intelligence, AI research chronology and AI research and discovery.

Public adviser — source-comparison laboratory

The first implementation helps a visitor document a public-interest question and compare two independent source descriptions. It requires original links to be supplied by the visitor. It reports differences and missing evidence; it does not declare an article true merely because several outlets repeat it.

Interactive prototype · local browser processing only

Enter two independently attributed source summaries and choose Compare. This demonstration checks record completeness and common vocabulary; it cannot verify factual claims or infer causality.
What an actual AI adviser would additionally need

Licensed or permissioned ingestion and source APIs; a dated claim-and-event store; primary-source retrieval; multilingual document parsing; recognition of independently sourced versus syndicated coverage; corroboration and contradiction assessment; explicit uncertainty and correction histories; transparent forecasts tested against outcomes; and editorial review for consequential conclusions.

Any public claim about an organisation, business or political event must distinguish confirmed facts from interpretations and scenarios. Political recommendations remain for users and accountable institutions to decide. No automated trading, hiring, benefit adjudication or government decision-making is implemented here.

Sources, independent scrutiny and responsible oversight

These organisations serve different functions: some produce evidence, others enforce obligations or independently review their field. None is a universal guarantor of truth. More sources can be added after provenance checks.

Organisation / sourceRoleRelevant question
UK Cabinet Office — National Risk Register 2026Official public risk assessmentWhich serious UK risks are identified, and what likelihood/impact assumptions are disclosed?
UK Be Ready campaignHousehold and community preparedness guidanceWhat practical, low-cost resilience steps are publicly recommended?
ACLEDPolitical violence and protest event dataWhich events are covered, how are they coded, and how are corrections handled?
UN OCHA — ReliefWebHumanitarian reports and response informationWhat is documented about affected populations and assistance needs?
GDELTGlobal event and news-derived knowledge representationsCan cross-language claims and events be compared without confusing media attention with incidence?
Reuters Institute — Digital News Report 2026Independent research on news access and trustHow do source diversity and public confidence affect AI-mediated information?
International Labour OrganizationResearch and international labour standardsHow do task exposures differ from proven job displacement?
Office for Budget ResponsibilityIndependent UK economic and fiscal forecastsWhich economic assumptions, scenarios and forecast revisions matter?
US NISTAI risk management guidanceHow are risks identified, evaluated and monitored?
UK Information Commissioner's OfficeData protection regulatorHow are privacy, fairness, data quality and individual rights protected?
Competition and Markets AuthorityUK competition and consumer protectionWhen do markets, market power or commercial claims warrant scrutiny?
OfcomUK communications and online-safety regulatorWhat obligations govern information platforms and safety?
News coverage register: compare, do not equate

Potential editorial viewpoints and reporting regions can be compared using BBC News, Al Jazeera, CNBC, Fox News, Reuters and independent or local reporting where relevant. Publisher variety does not automatically produce independent evidence: the same wire report may be republished by many outlets. Check primary documents, reporting dates, original evidence, corrections, publication rights and geographical coverage. No live articles are fetched by this page.

Economic opportunity and resilience beyond revenue

DimensionWhat the adviser should assessEvidence and safeguards
Enterprise opportunityDemand shifts, lawful market entry, supplier resilience, service gaps and funding optionsVerified business data, dated assumptions, uncertainty and failure cases; no promised returns
Continuity planningOperational dependencies, alternatives, workforce impact and recovery optionsSector-specific indicators, cost disclosure and permissioned external data
Employment transitionSkills transfer, training routes, accessible opportunities and work redesignILO and labour-market evidence; task exposure is not identical to actual unemployment
Non-market contributionsMutual aid, caregiving, community education, cooperative production and civic knowledgePractical participation without assuming everyone must monetise every activity
Household preparednessEssential communication, accessible emergency information and community supportLink to current government and local emergency guidance, not speculative danger predictions
Public-interest accessFreely available baseline intelligence and accessible explanationsSeparate any optional paid enterprise features from essential public-safety information

The ILO's 2025 task-exposure assessment and OECD Employment Outlook 2026 support scenario-based employment analysis, not a definitive claim that mass unemployment is inevitable. The UK Be Ready guidance supports practical preparation without alarmism.

Engineering, testing and potential product development

Stage 1 — Evidence-led demonstrator (available on this page)

Structured question capture, comparison of two source records, provenance completeness checks and an explicit statement of unsupported conclusions. Demonstration output is rule-based JavaScript and not AI inference.

Stage 2 — Controlled research assistant

Build a licensed/authorised retrieval backend, dated claim-to-evidence records, human-reviewed contradiction tests, web-source coverage reporting and correction/version histories. Compare against GDELT, existing research agents and manual research baselines.

Stage 3 — Calibrated public and enterprise advisory

Connect verified public datasets, business indicators, official risk registers and labour-market data; produce scenario assumptions and confidence limits; evaluate prediction calibration and potential harms before adding public advice features.

Stage 4 — Sustainable delivery and public access

Compare grants, research partnerships, enterprise subscriptions, responsible licensing, donations and cooperative structures. Maintain a freely accessible public-interest evidence layer, with editorial accountability and separation of financial incentives from factual conclusions.

Existing AI engineering roadmap · AI systems demonstrations · Research hub · Public productivity record

INTEGRATED SOCIETAL INTELLIGENCE · RESEARCH EXTENSION · NOVEMBER 2026

Moral formation, media environments and preventing escalation

Working hypothesis, not an established causal finding: how people are formed through families, faith communities, education, workplaces, media and civic institutions may affect their ability to resist dehumanisation, seek peaceful settlement, recognise foreseeable harm and exercise authority conscientiously. These pathways may interact; none alone explains a particular war.

Core research question: Under which conditions do sustained ethical formation, social connection, responsible information environments and accountable leadership reduce acceptance of coercion and improve peaceful responses to escalating disputes?

Use the following collapsible avenues as one shared research framework. Connected pages retain their own disciplinary context rather than reproducing the full investigation.

01 · Faith communities, discipleship and conscience formation

Context: examine sustained teaching, lived example, mentorship, care and peaceful coexistence in faith communities, while recognising that moral formation also occurs in secular settings. A claim that churches have ceased teaching discipleship requires denomination-, region- and time-specific evidence.

Potential contributions: curriculum and pastoral-practice reviews; voluntary community interviews; comparative case studies; safeguarding and accountability evaluations.

Question: Does consistent, voluntary moral formation correlate with compassionate conflict resolution and ethical conduct, after accounting for selection effects and socioeconomic context?

Falsification: no measurable improvement against comparable groups or outcomes better explained by other factors.

Faith and formation context · UNESCO education for peace recommendation (2023)

02 · Schools, families and lifelong civic education

Context: evaluate instruction and lived practice concerning self-control, non-coercion, empathy, critical reasoning, pluralism and appropriate boundaries; distinguish teaching principles from compulsory religious conformity.

Potential contributions: longitudinal school evaluations, peer mediation studies, family and community learning, teacher-developed materials.

Question: Which programmes improve observed dispute resolution, and for whom?

Falsification: benefits vanish in stronger controlled comparisons, do not persist or produce exclusionary effects.

UNESCO education recommendation

03 · Social media algorithms, journalism and war-related desensitisation

Context: investigate repetitive conflict exposure, hostile framing, misinformation, engagement optimisation, media literacy and legitimate discussion. Do not assume news exposure causes desensitisation or that lawful disagreement should be suppressed.

Potential contributions: exposure audits, consenting user-panel studies, editorial standards reviews, recommender-system experiments and source diversity checks.

Question: Do specific content-ranking patterns causally change empathy, dehumanising attitudes, accurate understanding or willingness to support peaceful settlement?

Falsification: no effect after controls, opposite effects, or intervention costs to expression exceeding established benefits.

UNESCO digital-platform guidelines (2023) · UN hate-speech prevention strategy (2019)

04 · Social connection, isolation and community resilience

Context: examine weakened social ties, loneliness, intergenerational support, exclusion and community-level trust without depicting isolated people as violent or morally deficient.

Potential contributions: community interviews, voluntary social-connection programmes, longitudinal surveys and local support mapping.

Question: When does meaningful cross-community contact improve trust and willingness to resolve differences peacefully?

Falsification: effects do not survive comparable-group analysis or are mediated entirely by unrelated variables.

WHO social connection overview (2025)

05 · Professional formation, institutional incentives and public-interest leadership

Context: investigate whether professional socialisation, advancement incentives and organisational norms reward conscientious challenge or discourage concern for affected people. Neither occupational background nor academic qualifications establishes moral character.

Potential contributions: documented decision-chain audits; interviews across professions; safe reporting-route reviews; counterfactual governance tests.

Question: Are documented ethical safeguards, challenge routes and independently audited decisions associated with fewer preventable harms?

Falsification: no effect, measurement bias or alternative explanations account for apparent improvement.

Institutional fairness and accountable leadership

06 · Migration, intercultural understanding and shared civic principles

Context: study how people with different experiences establish fair norms of coexistence, consent and mutual respect. Avoid treating migrants, religious groups or national communities as inherently threatening.

Potential contributions: intercultural dialogue pilots, civic-literacy evaluations, community mediation, rights-based policy comparison.

Question: Which interventions improve reciprocal trust without reducing freedom of conscience or equal rights?

Falsification: no improvement over comparable communities, or benefits limited to one group at the expense of another.

UNESCO global citizenship and peace education

07 · Crisis decisions, diplomacy and conflict early warning

Context: distinguish long-term formation from immediate policy and military decisions. Analyse warning signals, escalation pathways, available alternatives and accountability without diagnosing political leaders or attributing private motives.

Potential contributions: source-linked crisis timelines; matched escalating/de-escalating case studies; peacebuilding process comparisons and warning-to-response traceability.

Question: When credible warnings exist, which documented institutional responses reduce escalation?

Falsification: no comparative improvement, reverse causality or missing decision evidence.

UN–World Bank Pathways for Peace (2018)

08 · Evidence integrity, AI assurance and independent correction

Context: develop reproducible methods for identifying unsupported causal claims, inaccurate sources, suppressed warning records and miscalibrated AI-generated judgments. Human oversight must be qualified, impartial and contestable.

Potential contributions: provenance schemas; test fixtures; model evaluations; evidence retention and correction registers; independent QA protocols.

Question: Can provenance-based review improve early-warning credibility and reduce false attribution compared with ordinary review?

Falsification: no improvement on independent holdouts, excessive false positives, or evidence burdens greater than benefit.

Existing Global AI QA working preprint · Contextual research questions

09 · Evidence base, research boundaries and contribution opportunities

Established foundations: UNESCO's 2023 peace-education recommendation provides a rights-based educational framework; its 2023 digital-platform guidelines address governance of information systems; WHO's 2025 social-connection resource characterises social isolation; the UN–World Bank's 2018 work examines inclusive conflict prevention. None alone establishes that a particular conflict would have been easily prevented by a particular church or platform intervention.

Working roles: observers may raise questions; field researchers gather consented evidence; subject specialists assess methods; independent QA reviewers verify findings; collective outputs use explicit attribution, privacy and permission controls. Contributions are not automatically research publications or model-training material.

Measurement: use pre-registered outcomes, comparable populations, uncertainty intervals, negative-control tests, independent replication and clearly documented limitations. Separate correlations, credible causal evidence and normative proposals.

Primary references: UNESCO (2023), peace education; UNESCO (2023), platform governance; WHO (2025), social connection; UN–World Bank (2018), conflict prevention; UN (2019), hate-speech strategy.

Status: conceptual research extension. The proposed connections have not been tested, no individual or group is assigned blame, and there is no claim of proven causal prevention of specific wars. The shared framework may be corrected as evidence develops. Research index · Exploratory research horizon