The Transformer
Attention Is All You Need introduces the Transformer architecture.
Reading prompt: How did the architecture change the experimental baseline?
Google Research / Vaswani et al. — original publication ↗WORKING PREPRINT · VERSION 0.1 · NOVEMBER 2026
Public research in progress: 150 candidate assurance considerations, 15 pilot test designs and proposed continuous evaluation methods. Draft framework only; tests and independent validation outstanding.
RESEARCH AND PUBLIC CONTRIBUTION · NOVEMBER 2026
Open research agenda spanning 10 illustrative domains, 57 topic areas, model evaluation limitations and evidence-led questions that researchers and public contributors can investigate. The map is an exploratory taxonomy, not a claim about any person's full history or a tested model.
Concept, project, evidence and questions · Working paper protocol · Contextual question · Contribute evidence
CONCEPTUAL INSTITUTIONAL RESEARCH · NOVEMBER 2026
A contextual, evidence-led proposal for voluntary research cooperation across independently governed AI systems and institutions. No consortium membership or partnership is asserted.
Concept and project · Research question with context · Paper initiative · Log entry
AI SAFETY · RESEARCH INITIATIVE · 13 NOVEMBER 2026
A research protocol for testing whether AI filters can recognise their own errors, preserve potentially valuable knowledge, resist manipulation and avoid unsafe acceptance. Established research initiative; experimental findings are not yet claimed.
Research initiative and evidence register · Paper and experimental protocol · Dated log entry
One-page methodological references for existing AI, research, assurance and operations workstreams. Grey example snippets are non-executable, with plain-language explanations and evidence-status links. Browse seven project explanations.
Government tender portals, official research grant searches, partnership routes, institutional capability requirements and bid-readiness guidance. Open the funding and procurement gateway. Existing funding directories remain in place.
A beginner-friendly guide to distinguishing fluent summaries from explanatory depth, genuinely distinct possibilities, contextual understanding and verifiable completeness. Includes a transparent conversation observation table and proposed comparative testing protocol. Read the testing guide.
A research initiative connecting demographic unemployment differences to rents, energy, food systems, taxation, finance, asset ownership and ultimate distribution of income. Includes 2026 employment rates, 2024–25 household spending statistics, source-linked flow analysis and falsifiable questions. Read the systems research framework.
A 40-area exploratory framework for understanding organisational structures, authority, processes, policies, accessibility, evidence, oversight and preventive redesign. Distinguishes procedural departures, defective procedures, cross-system failures and unresolved evidence without publishing identifiable personal cases. Read the systemic review framework.
A searchable record of 15 documented enquiries arising from independent observation, professional experience, research and interdisciplinary questions. Entries distinguish open questions from linked initiatives and completed work, with evidence sources, international comparison paths and potential research-funding relevance. Browse the research questions and their source framework.
A research directory of 15 UK organisations connecting universities, employer demand, occupational standards, research funding, careers and professional accreditation. Includes official enquiry emails and contact-source links; distinguishes academic inquiry from business performance and professional recognition. Open the directory and contact table.
Preserves the 2024–25 approved specification and Year 2 dissertation structure while extending the study map with all ten original option modules, historic reading anchors, a 50-question scored self-assessment, 50 doctoral topic candidates and a separate 20-project learning-outcome register. Open the interactive preparation page.
Source-preserving 2024–25 MA programme and history syllabus register, full optional-module list, second-year dissertation architecture, 33-question self-scoring diagnostic across eleven knowledge areas, and doctoral thesis-development pathways. Open the interactive preparation lab.
Evidence-based review of professional critical-thinking certification and OfS university AI assessment regulation, with statistical context, potential institutional bottlenecks and falsifiable access questions. Read the reference flag (11 November 2026).
Evidence-bounded research chronology of relevant discussions from November 2024, further 2025 leads and the 2026 technical proposal, with a systematic initial prior-art comparison. Read the historical research record.
Author-originated research proposal investigating independent architecture discovery, quality and legal-compliance audits, structured assurance records and controlled system improvement. Includes an initial prior-work comparison. Read the research proposal.
Exploratory analysis of leading AI providers, including the July 2026 independent safety-index bar chart, quality questions, transparency gaps and the researcher\u2019s preliminary findings. View comparative AI assurance analysis.
An open-source Python scoring harness and twelve synthetic, author-labelled practice cases test evidence support, contradiction and insufficiency across negation, dates, quantities, causal claims and provenance. This implemented engineering artefact supplies a baseline and per-challenge scoring; it is not an independently validated benchmark or published model study. See the interactive systems demonstrations for related browser prototypes.
Explore a modular AI development programme spanning source-grounded research, business workflows, education, media scheduling, messaging and private knowledge retrieval. Test interactive prototypes and review the secure engineering roadmap. Public prototypes are local demonstrations rather than deployed autonomous agents.
AI research scientist training: Doctoral paper-writing skills and reproducible research standards.
Systems research: Completeness, proportional fairness, ecosystem stewardship and AI decision-support research (2026–2036).
Framework comparison: National & Global Frameworks Completeness Atlas — 33 frameworks and a proposed coverage audit.
Education research programme: Human capability, lifelong learning, apprenticeships and education policy change (2006–2026).
Browse by research question: AI governance, AI agents, fact-checking, digital rights, algorithmic fairness, higher education and research templates.
Practical research guides: How to verify AI-generated claims · How to audit AI bias and reliability · How to document non-traditional experience (CESI).
Research Collaborations — browse the topic library ↗
My research interests sit at the intersection of artificial intelligence, ethics, governance, technology, institutions and society, with particular attention to the gap between systems as designed and systems as experienced in practice.
Explore how AI could explain perceptual differences, visual and ecological harmony, the wider social value of beauty and the evidence needed for reflective, accountable decisions. Read the research proposal.
Independent scrutiny research: Explore 116 international oversight and accountability case-study candidates in a compact, searchable evidence directory. This is a preliminary screening register; individual findings have not all been independently audited.
A cross-government research programme covering the constitutional settlement, democratic institutions, public administration, fiscal responsibilities, services, justice, economic resilience and digital government. The directory maps 30 proposed research themes in five connected groups; individual topics are exploratory questions, not completed investigations or institutional findings.
Scope: Parliament and legislation; ministers and departments; the Civil Service; devolved administrations; local and regional authorities; regulators, inspectors and ombudsmen; public finance and contracted providers; courts and public services. This programme does not assume that one institution controls every function or that devolution arrangements are identical across the UK.
How to use this directory: Expand a group and follow the official starting source or the existing Manifold Grace page. Related links reuse earlier research rather than republishing it. Ordinary reading, scrolling, source inspection and browser search remain available without scripts.
Recorded 12 November 2026 · linked exploratory initiative, not a completed policy evaluation. Political commentary on immigration, fiscal constraints and the Church Appointments (Repeal of Discriminatory Provision) Bill raises testable questions about how government defines urgency, balances competing responsibilities and interprets longstanding legal restrictions. Commentary is treated as a research prompt rather than proof of causation or negligence.
Documentary context: The Parliamentary Bill page, House of Commons Library briefing (15 October 2026) and 10 November Lords debate describe proposals to repeal restrictions on Catholic and Jewish officeholders advising on church appointments. The Prime Minister's official record and Home Office small-boat arrival reports provide independent baselines; arrival counts alone cannot identify migrants' motivations or eventual immigration decisions.
Context: Governments may respond to urgent pressures and inherited legal anomalies at the same time. Compare parliamentary timetables, departmental capacity and policy deadlines; determine whether addressing one issue demonstrably displaced another, rather than inferring it from sequence alone.
Context: Church-state conventions and formal ministerial roles must be separated from doctrinal questions. Examine the specific statutory functions, the established appointment process and less restrictive alternatives before judging necessity or proportionality.
Context: Statements about arrivals, asylum outcomes, motivations and deterrence describe different phenomena. Use dated Home Office counts, decision statistics, policy commencement dates and robust causal comparisons; avoid assigning motives to people without evidence.
Context: Parliamentary attention is finite, but legislation also runs through concurrent departmental processes. Compare debate hours, staff and fiscal resources, decision delays and plausible counterfactual schedules rather than presuming a minor bill prevented action on other issues.
Context: Ministerial religious restrictions, the monarchy's religious settlement and church governance operate under different legal arrangements. Construct a consistent comparison of roles, rights, justifications and reform routes without treating them as legally identical.
Context: Speeches, budgets, enacted law, implementation and measurable effects are distinct stages. Trace claims to official records, oversight findings and timelines, recording contradictions, uncertainty and corrections.
Method: Gap-first historical and legal literature review; claim-to-evidence mapping; institutional process tracing; comparative policy analysis and falsifiable hypotheses. No findings of misconduct, policy causation or uniquely original contribution are asserted. Dated productivity-log record.
How are powers and duties distributed between Parliament, the executive, the Crown, courts and other public institutions?
Research direction: Institutional map · Official source / starting point
What happens between policy proposals, an Act receiving Royal Assent, commencement, implementation and independently verified outcomes?
Research direction: Legislative timeline · Official source / starting point
How do electoral systems, boundaries, participation, standards and representation affect democratic accountability?
Research direction: Comparative institutional review · Official source / starting point
How do powers, budgets, responsibilities and outcomes differ across the four UK nations?
Research direction: Four-nation comparison · Official source / starting point
How do councils, mayors, combined authorities and central departments coordinate public services?
Research direction: Responsibility map · Official source / starting point
Where do political decisions, administrative advice, delivery responsibilities and individual accountability intersect?
Research direction: Accountability pathway · Official source / starting point
How do tax design, spending, debt and financial forecasts affect long-term opportunity and service capacity?
Research direction: Fiscal effects comparison · Official source / starting point
How are publicly financed contracts, external suppliers, service outcomes and public-value obligations scrutinised?
Research direction: Contract-to-outcome audit · Official source / starting point
What makes a regulator independent, effective, accountable and accessible to those affected by its decisions?
Research direction: Regulatory map · Official source / starting point
How do recruitment, development, retention, leadership and working conditions shape public delivery?
Research direction: Capability analysis · Related Manifold Grace research
How are transport networks, buildings, digital infrastructure and public assets maintained across their lifecycles?
Research direction: Lifecycle assessment · Official source / starting point
How well do official statistics, evaluation designs and audits support policy decisions and corrections?
Research direction: Evidence assurance framework · Official source / starting point
How do funding, staffing, service design and regional conditions shape outcomes and access?
Research direction: Service performance comparison · Official source / starting point
How can health, housing, family support, safeguarding and financing coordinate across the lifespan?
Research direction: Care pathway mapping · Official source / starting point
How effectively can education and skills systems recognise capability and support progression?
Research direction: Access and progression study · Related Manifold Grace research
How do supply, tenure, land use, planning, infrastructure and tax interact with security and mobility?
Research direction: Intergenerational effects map · Related Manifold Grace research
How do eligibility rules, administrative delivery, reviews and appeals affect stability and access?
Research direction: Administrative journey audit · Official source / starting point
How do population ageing, early parenthood, unpaid care and households affect public policy?
Research direction: Joined-up family policy study · Related Manifold Grace research
How do investment choices, transport fares and accessibility affect participation and productivity?
Research direction: Access comparison · Official source / starting point
Which investments support durable employment, innovation, productive capacity and regional growth?
Research direction: Policy portfolio analysis · Official source / starting point
How do employment conditions, contracts, pay structures, recruitment and outsourcing interact?
Research direction: Labour-market interaction map · Related Manifold Grace research
Can people obtain dependable evidence, timely remedies and fair review across the justice system?
Research direction: Remedy and evidence framework · Related Manifold Grace research
How do legal rights, migration administration, integration, population change and public services intersect?
Research direction: Institutional comparison · Official source / starting point
How are affordability, climate commitments, environmental quality and energy security weighed?
Research direction: Trade-off assessment · Official source / starting point
How do farming, food systems, nature, land stewardship and rural public services influence resilience?
Research direction: Food-system resilience map · Official source / starting point
How do diplomatic commitments, defence capabilities, national security and public spending connect?
Research direction: International obligation register · Official source / starting point
Why do public bodies struggle to exchange accurate information and provide coordinated services?
Research direction: Interoperability assessment · Related Manifold Grace research
How can algorithm-assisted public decisions be explained, reviewed, corrected and independently audited?
Research direction: Appeals and audit model · Related Manifold Grace research
Do complaints, review processes and appeals produce sustained improvement beyond individual case closure?
Research direction: Learning-and-remedy tracker · Related Manifold Grace research
How can costs, benefits, distributional effects and unintended consequences be evaluated across departments?
Research direction: Cross-system impact model · Related Manifold Grace research
For each topic, trace legal authority → responsible institutions → funding and delivery → measured outcomes → independent scrutiny → remedies and reforms. Distinguish confirmed powers and published facts from hypotheses; compare devolved jurisdictions before drawing UK-wide conclusions. Cross-department effects should be studied, not presumed.
Foundational sources: How government works (GOV.UK) · Government departments, agencies and public bodies · UK Parliament's role · National Audit Office. Source links are starting points; their presence does not independently verify every research question.
Research how searchable handwriting, linked archive descriptions, transparent evidence graphs and interoperable genealogy formats could help future generations. Explore historical corridors and AI recordkeeping · Dated source log.
Evidence-led research on institutional access, progression, economic security, distribution of risk and accountability, with reform hypotheses and evaluation measures. Read the five-test framework and linked sources · Research log.
A developing three-layer research architecture for knowledge mapping, evidence evaluation and historical systems analysis, with milestone tracking and a proposed 2007–2027 UK policy application.
Navigate connections among academic programme themes, research domains, methods, industries and public institutions. The initial crosswalk distinguishes thematic affinity from assessed learning and is designed for expansion beyond the uploaded 2024–25 specifications.
A concise index of 38 research arenas, with AI safety, security and assurance prioritised for current development; the wider horizon remains open for later work.
Browse the research arenas → My MA-connected research avenues → Bottleneck project avenues →RELATED INTERDISCIPLINARY RESEARCH · 3 NOVEMBER 2026
Evidence Verification Across Sectors — research and applications →
Source provenance, historical records, claim verification and AI evaluation for education, RPL, teaching, healthcare, legal practice, accounting, audit and public administration. Each field retains its own standards and professional judgments.
Separate specialist study: criminal justice evidence evaluation →
Associated Press reports that a Pakistani astronaut is expected to fly aboard China's Shenzhou-24 mission, becoming the first foreign national to visit the Tiangong space station. The announcement offers a documented example of international access to space research infrastructure and raises comparative questions about scientific partnerships, research access and governance; it does not alone demonstrate equitable access across countries.
Source: Associated Press — Pakistani astronaut and Shenzhou-24
An interdisciplinary inquiry into how social experience, statistics, history, philosophy, reporting and technological change can inform music, visual art, fashion and cultural expression—while preserving clear distinctions between evidence, interpretation and imagination.
Risk considerations are documented for each of the 63 free and low-cost AI tools, with 20 historically documented security and governance case studies. Coverage distinguishes incidents from unproven allegations, and uses technical and jurisdictional evidence rather than assumptions about staff identities.
Reviewed 23 Salesforce, Agentforce, CRM Analytics and open-tool integration pathways, distinguishing genuinely free developer and small-business options from paid analytics and metered production services. Browse the enterprise scaling and integration register.
Registered a source-aware visual-knowledge audit and organisation-discovery project with comparisons against established research, measurable proposed experiments and a designated paper initiative. No unlicensed archive extraction or completed findings are claimed.
A cost-first technology register covering model runtimes, hosted free allocations, evidence retrieval, databases, evaluation and workflows, with account requirements, licence restrictions, alternatives and validation tasks. The expanded inventory contains 63 candidates and distinguishes actual integrations from proposals.
Browse the searchable AI infrastructure and integration register.
Interdisciplinary research into geopolitical uncertainty, political artificial advisers, probability, data value, state power, explainability, employment transitions and accountable AI. A first browser-only demonstration supports documenting two public-source claims and identifying what cannot yet be verified; it is not a live news AI or predictor.
Comparative AI research history: 22 selected humanitarian, public-data, computational event, knowledge-graph and global-monitoring systems, with founding or launch dates and original sources. Related to the history and foundations of AI studied at master's level and subsequent independent research; it is not a claim of completed academic assessment.
Explore the dated platform chronology and prior-art questions. The comparison distinguishes historical achievements from candidate new research contributions, including missing evidence, contradiction detection and unknown relationships.
Selected primary publications, arranged by year. A reading trail for returning to important ideas, rather than an expectation to remember every article. The ten-year anniversary of 2017 falls in 2027.
Curated historical highlights · checked 9 October 2026 · automatically scrolls at a steady pace while open. Hover or focus to read; swipe to explore. Reduced-motion preferences are respected.
Attention Is All You Need introduces the Transformer architecture.
Reading prompt: How did the architecture change the experimental baseline?
Google Research / Vaswani et al. — original publication ↗Google announces open-sourcing BERT for language representation.
Reading prompt: What did pre-training change about task adaptation?
Google Research — original publication ↗OpenAI releases the 1.5B-parameter GPT-2 model after a staged release.
Reading prompt: What evidence should guide publication and misuse decisions?
OpenAI — original publication ↗OpenAI publishes Language models are few-shot learners.
Reading prompt: When does few-shot performance generalise beyond a benchmark?
OpenAI — original publication ↗DeepMind describes AlphaFold’s method, open-source code and database release.
Reading prompt: What makes a prediction useful to independent scientists?
Google DeepMind — original publication ↗OpenAI introduces ChatGPT and invites feedback on strengths and weaknesses.
Reading prompt: Which failure modes emerge when research reaches public users?
OpenAI — original publication ↗OpenAI publishes its GPT-4 research announcement.
Reading prompt: Which claims are reproducible and which remain provider-reported?
OpenAI — original publication ↗Google announces AlphaFold 3, developed by DeepMind and Isomorphic Labs.
Reading prompt: How should scientific scope and downstream validation be separated?
Google — original publication ↗The Gemini Team documents reasoning, multimodality and long-context capabilities.
Reading prompt: What do capability gains cost, and where do they fail?
Google / Gemini Team — original publication ↗Anthropic’s Cadences report examines Claude usage and initial user-survey findings.
Reading prompt: How representative is one provider’s usage data?
Anthropic — original publication ↗I follow developments through articles, research releases, email updates and professional networks, including LinkedIn. I return to sources when a question becomes relevant; keeping informed does not require memorising a decade of articles.
I value the quality of a contribution, the problem it solves, the change it enables and the questions it opens. My publishing pace needs to make room for postgraduate research, practical exploration, professional commitments and life.
Working practice: trace an update to its original source; separate a summary from an original finding; publish a short, attributed note when useful; develop a substantial paper when the method and evidence are ready. There is no daily publication commitment here.
This timeline records public developments. Dated personal notes, saved articles and correspondence can document an individual reading history.
My research is deliberately interdisciplinary. It connects questions that are often separated institutionally — technical design, law and regulation, organisational behaviour, public administration, education, ethics and social consequences.
The aim is not simply to describe emerging technologies, but to examine how they are governed, implemented and experienced, including the points where formal policy and operational reality diverge.
Preparation for future doctoral research has also been in formation through institutional pathways, including within the University of London context. This remains preparatory research development rather than a claim of current doctoral candidacy.
Search arXiv, the open-access research preprint archive, for work in computing, mathematics, physics and related disciplines. Organise papers you want to read with an expandable reference list and linked original sources.
Choose an area or enter a query.
The results open on arXiv rather than within an iframe: third-party embedding and browser-only live API access cannot be relied upon on a static GitHub Pages website. This provides direct access to current results without misrepresenting an external feed as embedded.
After opening a paper on arXiv, save its actual title, source URL and reading notes here. Only papers you add appear in the catalogue; no papers or publication dates are invented.
The catalogue is saved in this browser only.
Browser storage is device-specific, can be cleared and is not cloud synchronised. Export regularly. For a public bibliography, use checked references and suitable licensing.
Many arXiv records are preprints and should not automatically be described as peer reviewed. Check the posted version, later journal publication, methods, data, conflicts of interest and independent follow-up. Cite the actual authors and paper version, and distinguish proposed findings from established results.
Explore representative doctoral-level workflows, from gathering evidence to running reproducible experiments. Each example includes the research task, practical steps, intended output and links to established tools. These are proposed starter projects, not claims of completed experiments or connected laboratory services.
Open interactive project & annotated code →
Identify themes and research gaps across a scholarly literature corpus using NLP, while keeping researcher review central.
Output: Evidence map, bibliography and auditable screening log.
Open interactive project & annotated code →
Build a trustworthy dataset for scientific AI work, with explicit permissions, quality checks and traceable transformations.
Output: Versioned dataset, data dictionary and provenance record.
Explore: DVC · Label Studio · Jupyter
Open interactive project & annotated code →
Compare a defensible baseline against alternative machine-learning models, with reproducible evaluation.
Output: Results table, experiment log and reproducible notebook.
Explore: scikit-learn · MLflow · Colab
Open interactive project & annotated code →
Investigate where an AI model performs inconsistently and document governance implications using measurable tests.
Output: Evaluation report, model card and risk register.
Explore: NIST AI RMF · SHAP · Model cards
Open interactive project & annotated code →
Use computational models to explore complex systems, uncertainty and the consequences of changing assumptions.
Output: Reproducible simulation, uncertainty plots and methods note.
Swipe sideways on touchscreens, scroll horizontally, or use the arrows to explore all five. The named services are external tools; secure private research execution would require a separately hosted backend.
Launch the interactive AI Research Laboratory → · Five runnable demos, code walkthroughs, security annotations and advanced research questions.
The wider research on AI governance, institutional histories and source provenance has a possible application to evaluating evidential accuracy in criminal justice. Documented claims could be traced to original records, dated chronologies reviewed for inconsistencies, missing sources identified, and historical policies compared with the versions applicable at the time. The aim is to make evidential uncertainty and provenance visible to authorised human reviewers, whether evidence supports or challenges an allegation.
This is one possible use among education, employment, healthcare, research and public administration; it does not redefine Manifold Grace as a legal practice or claim a deployed legal tool. Potential implementations require lawful handling of sensitive information, secure access and qualified professional review.
Information-science design · Evaluation tests · Dated contribution context.
The research area is divided by function rather than by prestige: reference links for navigation, case studies for bounded problems, papers for longer outputs, and Systems for questions that only make sense when institutions are mapped together.
This keeps working material, published material and source infrastructure separate while allowing them to connect.
These profiles connect my research identity, scholarship, repositories and public writing. The links below include the direct addresses supplied for this record.
Profiles, developing work, research methods and source collections brought together across my areas of inquiry. Open a category to browse its links.
Working papers are labelled as drafts. Follow the research profiles to find authored work, or browse the thematic categories for datasets and other evidence sources.
Reference resources for cross-checking pedigrees, primary records, estate succession, historical biographies, cultural conservation and current policy. Compiled family trees are discovery aids, not independent proof of descent. Sensitive living-person family information is not published here.
A working glossary connecting my research interests across ethics, AI, institutions, social experience and engineering. Definitions are concise orientation notes. Working concepts and research propositions are labelled to distinguish them from established terminology.
107 terms across 6 themes
No matching terms. Try a broader word.
Within the AI, Ethics & Society cohort, I contributed informal peer mentoring, conceptual-development support, curriculum engagement and knowledge exchange. I was invited to contribute a segment of a class session, drawing on professional experience in AI ethics, reflective methodology and socio-technical systems. I developed and shared the Chronological Ethical Self-Inventory as a reflective framework for ethical reasoning and personal knowledge management.
I have liaised with the Social Research Association and engaged with external academic networks, alongside voluntary representation and inclusion advocacy within my postgraduate community.
I submitted proposals on recognising participatory research contributions, preserving attribution and documenting evolving forms of AI-assisted scholarly work.
I submitted feature proposals concerning conversational memory transparency, accessible knowledge management, attribution, profile synchronisation and recognition of human contributions to socio-technical systems.
AI Product Feedback — selected proposals and improvement requests →
I documented institutional processes, prepared correspondence and appeals, and examined procedural fairness, data governance and accessibility. This work connects lived institutional experience with questions about research progression and accountable public systems.
My Chronological Ethical Self-Inventory (CESI) connects life-course reflection with evidence, ethical interpretation and applied strategy. A central question in this application is how institutions recognise relevant knowledge and contribution when these develop across different sectors, independent projects, education and periods outside paid employment.
Relevant observations deserve consideration across socioeconomic positions, educational backgrounds, sexes, gender identities, cultures and lived experiences. Identity or status alone does not determine the accuracy of a claim. Experiences and feelings can identify questions worth investigating; documented circumstances, contextual factors, alternative explanations and proportionate verification help establish what can be concluded. Inclusion broadens inquiry without requiring every interpretation to carry equal evidential weight.
These dimensions can diverge. A well-reasoned assessment can demonstrate capability even when its recommendation is not adopted. Adoption alone does not establish that a decision was correct or beneficial. Institutional change requires separate evidence before it can be attributed to an individual's contribution.
On 29 October 2026, I used an AI-assisted comparison of 20 recruitment adverts to examine stated requirements across governance operations, privacy, data governance, consulting and technical oversight. The purposive sample contained 15 UK roles, one international remote role and four overseas benchmarks. It included employer adverts and platform copies, with some closed or indexed records. It is a requirements comparison, not a representative vacancy survey or a study of actual selection decisions.
The comparison identified recurring communication, implementation and risk-related responsibilities, alongside differing technical and qualification requirements. It provides a practical context for asking how experience becomes recognisable evidence. It does not establish that a portfolio secures employment, that every candidate meets the criteria or that employers' actual decisions match their stated requirements.
On Manifold Grace, the activity log records actions and outputs; the evidence index sets boundaries for stronger claims; professional case studies explain the decisions and their relevance. A blank date records missing documentation, not inactivity. An AI-assisted reconstruction is a derivative account, not independent corroboration of the underlying event. This is a methodological application of a conceptual framework, rather than an empirically validated assessment of employability.
CESI preprint and publication record · Professional capability map · Activity log · Evidence maturity standard
Application commentary updated . The linked CESI preprint remains a separate publication.
I receive enquiries through email and professional networks about potential research, editorial and collaborative opportunities. I am exploring suitable connections, including ways my work on AI ethics, governance and accessibility could develop across disciplines. These conversations remain exploratory, and I am optimistic about the possibilities for meaningful contribution and professional progression.
— Policy research and parliamentary correspondence considered an optional 2–5-year parental career break, employment and pension protections, the existing NHS career-break precedent, and early-childhood safeguarding. An alternative parent-integrated workplace concept places staffed childcare beside suitable work or study areas to support continued professional development. These are proposals requiring feasibility, equality, safeguarding and cost evaluation, not implemented programmes.
Professional context: This work contributes to social governance and public-policy research; it does not indicate a return to childcare employment, as the author's professional direction is research, governance and policy.
Contribution and correspondence record · 4 November activity.
The proposed verifiable advanced intelligence research series links multimodal evidence synthesis, causal and stochastic reasoning, thirty domain-specific architectures and a controlled evaluation protocol. Read the proposed research framework.
Additive research note, 4 November 2026. Conceptual and testable; not a validated ASI capability.
An exploratory log records questions about mathematical proof correspondence, versioning, attribution, scientific verification and independent assurance, without treating proposals as completed research. View the research review log.
Logged 5 November 2026 · Exploratory · Follow-up investigation pending.
A dated reading crosswalk records historically grounded connections between feedback, communication, statistical prediction, learning and contemporary AI evaluation and governance questions. Read the dated course-reading crosswalk.
5 November 2026 · Source-based reading links and later research questions; no existing content replaced.
This research programme uses revisable landscape mapping to identify boundaries, relationships and competing perspectives before targeted evidence work, project design and contextual teaching. The full method explains how CESI may contribute where chronology, capability and recognition are relevant, without treating it as a universal framework. Read the methodology and project-mapping statement.
Methodological development dated 4 November 2026; this section supplements rather than revises existing records.
A new uncertainty-aware roadmap connects ongoing research with bounded deliverables, source pages and a small synthetic claim-analysis test pack. Original publication and feature records remain unchanged. Open the active research roadmap.
Research planning note · 5 November 2026 · Provisional priorities, not completed experimental results.
Inclusive, evidence-led civil-rights and equitable-living directory covering UK and US organisations, institutional accountability, practical initiatives and testable research questions.
Cross-industry, two-level collapsible directories for systems mapping, architecture comparison, tool inventories, compliance and ethical evidence, and agent-assisted investigation.
Explore the 20-industry systems atlas and investigation framework
Integrated seven-domain AI taxonomy, existing twenty-industry architecture atlas and twenty-workstream evidence register; a functional comparison demonstration and detailed agent validation stages are available for inspection. Production research agents remain subject to separate validation.
Explore the AI Knowledge Atlas · Test the architecture comparison prototype · Review 20 workstreams and engineering sequence
Public, static-first test tools and source templates for comparative systems, evidence verification and commercial opportunity assessment. The demos use sample data and do not provide legal certification or autonomous research.
Open interactive demonstration portal · View dated technology-market evidence · View twenty-workstream status and validation sequence · View source schemas, tests and templates
This distinct proposed research direction examines whether AI-assisted media can manufacture authority, spread unsupported claims and manipulate decisions. An advertisement need not be proven AI-generated to be misleading. The initial example concerns purported dementia treatments; the approach is intended to remain applicable across sectors.
Research status: Questions recorded for future investigation, not claimed as novel discoveries or validated findings. Scope and originality require a documented literature and prior-art audit. Related programme: Independent Research Questions Observatory and the existing evidence-verification work.
Initial evidence and regulatory anchors: NHS — Alzheimer's disease · FDA — False Alzheimer's cure claims · UK Advertising Standards Authority · Alzheimer's Society. These resources are starting points, not a completed systematic review.
Exploratory branch on how wartime knowledge gaps arise, with 12 future-team questions and ethical evidence-preservation measures for conflicts affecting people in 2026.
Recorded 12 November 2026 · Research initiative / exploratory questions, not completed findings. This independent line of enquiry examines whether AI adoption, market concentration and recruitment practices affect access to paid work and the distribution of economic benefits and burdens. Social research can provide evidence about experiences of exclusion and their institutional context, while economic and technical evidence tests causal mechanisms.
Combine longitudinal labour-market indicators, firm-level recruitment practices, AI-task exposure measures, anonymised qualitative interviews, distributive-impact assessment and competing causal hypotheses. Preserve the distinction between documented experience, reported allegations, market structure, correlation and demonstrated causation. Social-media commentary is an example of public discourse, not proof of AI causation or unfair conduct by any named party.
ILO (2025), Generative AI and jobs: A 2025 update — identifies occupational exposure while stressing that exposure is not equivalent to redundancy. ILO Working Paper 140 (2025) — refined task-level exposure measurement. These are starting points for a gap-first prior-art and falsification audit; no originality claim is made here.
Cross-links: UK institutions and public systems (where available); Dated research log.
Research record published 13 November 2026. This investigation extends the Manifold Grace programme on claim-to-evidence verification, provenance, independent evaluation and correction history. The proposed study connects versioned source changes, build provenance, release approvals, independently observed production behaviour and measured outcomes. Its hypotheses remain open to testing.
| Checkpoint | Verifiable record | Comparison and outcome |
|---|---|---|
| Development | Commit SHA, authorisation, pull-request review, source diff, dependency lockfile | Intended changes and documented pre-change baseline |
| CI and build | Test reports, security checks, reproducible build hash, signed provenance | Expected properties, known limitations and failed checks |
| Deployment | Release timestamp, environment, artifact digest and approval record | Confirmation that approved source maps to the deployed artifact |
| Operation | Independent smoke tests, API responses, accessibility, errors and performance | Observed behaviour contrasted with release claims and baseline |
| Longitudinal review | Incident timelines, corrections, rollback records, repeat measurements | Persistence of benefits, regressions, unexplained differences and alternative causes |
Evaluation standard: record observed changes without automatically inferring intent, wrongdoing or causation. Use controlled experiments or credible comparative designs to estimate effects. Quantified probabilities require validated reference data and stated assumptions; otherwise retain uncertainty explicitly. Internal source or system access requires proper authorisation, confidentiality and privacy safeguards.
The dates below are the original publication or release dates stated by the issuing organisations, not the date of this Manifold Grace entry. Later revisions, where known, must be shown separately rather than substituted for the original date.
| Original publication date | Source and release | Relevance and scope |
|---|---|---|
| 19 April 2023 | OpenSSF / Linux Foundation — SLSA v1.0 release announcement | Build integrity, software-supply-chain traceability and provenance standards. |
| 11 April 2024 | UK Competition and Markets Authority — AI foundation-model market concerns | Published regulator perspectives on market relationships and potential competition risks; not proof of unlawful coordination. |
| 2 May 2024 | GitHub — Artifact Attestations public beta announcement | Initial product availability; the page also carries a later update date, which is not the original publication date. |
| 25 June 2024 | GitHub — Artifact Attestations general availability | Signed build attestations and verification before deployment. |
| 17 January 2025 | US Federal Trade Commission — AI partnerships and investments staff report | Evidence of reported partnership structures and the FTC's assessment of potential competition implications; not a finding of illegal cartel activity. |
Record integrity: release dates, later update dates, dates on which evidence was collected, and the Manifold Grace publication date describe different events and should remain distinguishable. A publisher-provided timestamp is a claim about publication history; independent archive, repository or filing records can strengthen chronological verification.
Related research: Information Science · Research Questions Observatory · Dated research log.
Independent, continuously versioned index of Bloomberg AI and technology story metadata, with original publication dates, authors when supplied, direct publisher links and topic-based connections to the research programme. The archive distinguishes collection time from publication time, preserves prior entries across successful update runs and does not reproduce paywalled articles.
Open the Bloomberg AI Research Index · View collection workflow and run status
Structured educational and research pathways link original research, upstream code provenance and permission, beginner-to-advanced examples, compliance and safeguarding review, CI/CD evidence, deployed outcomes and distinct academic paper types.
Open the linked project and publication framework · Existing project register
A dated, version-controlled publication-history demonstration links daily, weekly, monthly and yearly PDF directories with searchable original-record metadata. Related working manuscripts explore archival transparency standards, lawful indexing, reproducibility, accessibility, higher-education curricula and possible AI evaluation benefits. Proposed regulatory obligations remain distinct from existing law.
Open the downloadable archive · Read architecture, policy and education research initiatives
A cross-domain research inquiry examines whether economic and policy barriers to children's educational and cultural participation contribute to differences in knowledge documentation and, through separate data-selection mechanisms, possible omissions or distortions in AI datasets and model evaluations. These causal links remain hypotheses, not completed empirical findings. The study includes local learning alternatives, privacy and safeguarding, and the potential benefits of reinvested visitor-levy funds.
Read the six-stage causal framework, falsification plan and primary sources.
Context recorded 13 November 2026. Accountability, apology, repentance, restitution and reconciliation can be explored within Christian theology, other religious traditions, Indigenous and community justice practices, philosophical ethics and secular legal frameworks. These are overlapping concerns, not exclusive Christian inventions. Differences in sources, aims, duties, authority, dispute processes and remedies must be preserved rather than assuming traditions are interchangeable.
| Source or perspective | What it contributes to comparative study | Qualification |
|---|---|---|
| Biblical passages | Acknowledgement, repentance, reconciliation and restitution. Read passages in context. | Interpret each verse within its historical and theological setting. |
| Jewish ethical and legal traditions | Study teshuvah, making amends, justice and obligations to other people. | Interpretations vary across periods and communities; avoid reducing these traditions to a Christian parallel. |
| Islamic ethical and legal traditions | Study tawbah, responsibility, forgiveness, restoration of others' rights and restitution where appropriate. | Specific religious and legal doctrines should be reviewed from primary sources and qualified scholarship. |
| Community and Indigenous justice approaches | Examine practices of accountability, repair, participation and restoring relationships. | These are distinct traditions, not a single universal Indigenous model. |
| Secular restorative justice and human rights | Compare acknowledgement, access to remedy, restitution, compensation, rehabilitation, satisfaction and guarantees of non-repetition. | Remedies depend on applicable law, harm and due process; not every interpersonal dispute is a human-rights violation. |
Verified secular framework: UN General Assembly Resolution 60/147 (adopted 16 December 2005) describes remedies and reparation for gross violations of international human rights law and serious violations of international humanitarian law, including restitution, compensation, rehabilitation, satisfaction and guarantees of non-repetition. Its precise scope should not be extended to every wrongdoing automatically.
Research questions: Which principles genuinely converge across traditions? Where do they differ in responsibility and remedies? How can acknowledgment, repair, legal due process and prevention of recurrence be operationalised in public institutions without using one faith as the exclusive moral authority? What evidence would distinguish sincere institutional correction from symbolic apology?
Status: comparative research direction; the table is a starting taxonomy and not an exhaustive literature review or completed cross-cultural study. Equitable Living · Scripture context · Dated log.
Browse a searchable table of 1,000 proposed AI configurations constructed from 40 base architecture families. Every architecture-family name is linked in blue to a relevant Wikipedia contextual article; the source table separately documents relational mechanisms, specialised components and untested research status. A second section compares shared relational frameworks, with a separate specialised-component matrix.
Explore the architecture atlas and specialised component matrix · Full CSV
Scrollable architecture atlas (embedded below). The table supports vertical and horizontal scrolling, filters and pagination; a download is not required.
CONNECTED RESEARCH · NOVEMBER 2026
Linked extension to existing societal resilience, faith, institutional fairness, research questions and AI assurance: eight investigation avenues with proposed contributions, measures, falsification criteria and primary sources. These are research hypotheses, not findings that a specific conflict was preventable.
Full context and source links · Formation and faith · Institutional stewardship · Research questions · Research horizon.