Ethics of political AI advisers
Human authority, accountable recommendations, institutional power, democratic scrutiny and conflict-sensitive information.
Research & Knowledge · Public Interest AI · Enterprise
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.
Human authority, accountable recommendations, institutional power, democratic scrutiny and conflict-sensitive information.
Calibration, baselines, scenario dependence, missing data, uncertainty and the difference between conditional forecasts and certainty.
Data provenance, unequal access, licensing, physical compute costs, infrastructure and environmental consequences.
Who is represented, who is absent, which assumptions drive outputs and how decisions can be challenged or corrected.
Conflicts, trade disruptions, critical services, public institutions, supply chains, resilience and humanitarian consequences.
Labour transitions, cooperative ventures, viable businesses, public goods and meaningful activity outside paid employment.
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.
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
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.
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 / source | Role | Relevant question |
|---|---|---|
| UK Cabinet Office — National Risk Register 2026 | Official public risk assessment | Which serious UK risks are identified, and what likelihood/impact assumptions are disclosed? |
| UK Be Ready campaign | Household and community preparedness guidance | What practical, low-cost resilience steps are publicly recommended? |
| ACLED | Political violence and protest event data | Which events are covered, how are they coded, and how are corrections handled? |
| UN OCHA — ReliefWeb | Humanitarian reports and response information | What is documented about affected populations and assistance needs? |
| GDELT | Global event and news-derived knowledge representations | Can cross-language claims and events be compared without confusing media attention with incidence? |
| Reuters Institute — Digital News Report 2026 | Independent research on news access and trust | How do source diversity and public confidence affect AI-mediated information? |
| International Labour Organization | Research and international labour standards | How do task exposures differ from proven job displacement? |
| Office for Budget Responsibility | Independent UK economic and fiscal forecasts | Which economic assumptions, scenarios and forecast revisions matter? |
| US NIST | AI risk management guidance | How are risks identified, evaluated and monitored? |
| UK Information Commissioner's Office | Data protection regulator | How are privacy, fairness, data quality and individual rights protected? |
| Competition and Markets Authority | UK competition and consumer protection | When do markets, market power or commercial claims warrant scrutiny? |
| Ofcom | UK communications and online-safety regulator | What obligations govern information platforms and safety? |
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.
| Dimension | What the adviser should assess | Evidence and safeguards |
|---|---|---|
| Enterprise opportunity | Demand shifts, lawful market entry, supplier resilience, service gaps and funding options | Verified business data, dated assumptions, uncertainty and failure cases; no promised returns |
| Continuity planning | Operational dependencies, alternatives, workforce impact and recovery options | Sector-specific indicators, cost disclosure and permissioned external data |
| Employment transition | Skills transfer, training routes, accessible opportunities and work redesign | ILO and labour-market evidence; task exposure is not identical to actual unemployment |
| Non-market contributions | Mutual aid, caregiving, community education, cooperative production and civic knowledge | Practical participation without assuming everyone must monetise every activity |
| Household preparedness | Essential communication, accessible emergency information and community support | Link to current government and local emergency guidance, not speculative danger predictions |
| Public-interest access | Freely available baseline intelligence and accessible explanations | Separate 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.
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.
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.
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.
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
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.
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)
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.
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)
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.
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.
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.
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.
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
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