03 — Representing genuine human diversity in AI training and evaluation

2026 research edition · Research topic 3/10 · CognitiveLiberties.com

Editorial position and scope

Proposed mission: Cognitive liberty means that people can think and inquire freely, seek and receive information, voluntarily share their views and evidence, encounter disagreement, and challenge opaque decisions that prevent lawful information from being found. Access does not equal endorsement: freedom to inspect a claim is compatible with labeling its evidence weak, correcting falsehoods, and applying proportionate lawful restrictions. Individual privacy remains an option and a safeguard, not the central objective or a demand that people stop sharing with knowledge systems.

Status of this document: Independent desk research as of October 11, 2026. It uses public sources and the existing public website. It is not an internal site audit, a measured search-ranking experiment, an inspection of proprietary AI training datasets, or legal advice. Numbered [S##] citations resolve to SOURCES.md.

Research question

How can more voices inform AI while maintaining provenance, community participation, respect for creators, and epistemic reliability?

Executive finding

Diversity is both a dataset coverage problem and a model-behavior problem. Original human writing from less represented languages and perspectives is worth preserving, but publishing a viewpoint does not ensure its inclusion in a training run. A contributor-governed knowledge commons and targeted evaluation provide more controllable interventions than simply allowing unlimited scraping.

Evidence from public sources

1. UNESCO calls for inclusive AI teams, training datasets and fair access, with specific attention to marginalized and multilingual communities. [S02][S06]
2. UNESCO’s multilingual roadmap identifies language technologies as part of broadening digital representation. [S25]
3. Poole-Dayan and colleagues reported in July 2026 that their audit of publicly available frontier-lab materials found no clear evidence that pluralism was a named training/evaluation objective in production systems. This is limited by publicly disclosed evidence. [S03]
4. Shumailov et al. studied degradation from recursively generated training data. Their “model collapse” result does not establish that suppressing unpopular ideas causes the same technical phenomenon. [S05]

Analysis: what these findings actually mean

1. The AI pipeline has at least six selection points: which people publish; what is crawlable; how content is licensed; how text is filtered and weighted; whose preference judgments count; and which evaluations a product must pass. Diversity can be lost at any stage, including after pretraining.

2. The highest-impact intervention may be evaluation, not simply dataset growth. If models fail to recognize minority philosophies or confidently misattribute dissenting claims, targeted evaluation can expose those errors even when the training corpus is undisclosed.

3. A knowledge commons should label contributions as eyewitness account, opinion, scholarly claim, research dataset, artistic expression or proposed hypothesis. Preserving dissent does not require laundering conjecture into fact.

4. Contribution permissions should distinguish searchable public display, retrieval and quotation, research dataset release, and training reuse. These are different decisions and may involve different legal or contractual terms. Contributor voice grows stronger when attribution and correction mechanisms exist.

Application to CognitiveLiberties.com

The goal is to make a meaningful discovery path between a visitor's curiosity, multiple independently operated sources, original evidence, fair representation of dissent and an explanation of how a source was selected. This is not merely a traffic-acquisition campaign. The current site includes usable public navigation and source records, including its Resource Navigator, network and evidence ledger. It should not be described as having broken endpoints based on a past unverified report. [S18][S19][S20][S21]

1. Action 1: Design an openly readable contributor registry with stable IDs, language tags, viewpoint descriptions and provenance links.
2. Action 2: Pilot five multilingual and multi-perspective topic collections, with independent editors and a published inclusion policy.
3. Action 3: Publish reproducible AI tests for accurate description of viewpoints, not merely agreement with them.

How to test whether the idea works

Reproducible test design

Unit: a documented question → source list → human interpretation task, not a single isolated search screenshot. Start with a preregistered set of ten to thirty representative tasks, including questions where expert consensus is strong and others involving substantial interpretive dispute. Keep question phrasing, language, device, time, geography, tools and source eligibility criteria recorded. Include at least two independently maintained competing sources where an actual debate exists, and assign evidence-strength labels through blinded review.

Baseline: test the current public site's navigation and at least one ordinary general-purpose search/AI route. Compare to a link-first trail and, only if built, an AI-assisted source mapper. Log whether a participant reaches an original source, recognizes disagreements accurately and can find a rebuttal without re-querying a centralized intermediary. Audit the alternative for false balance and omitted mainstream evidence as carefully as for omission of minority perspectives.

Interpretation: report descriptive rates, uncertainty and failure cases. Avoid causal claims from observational cross-platform comparisons. Record opt-outs and consent for human research; do not infer ideological categories from users' queries.

Measurements specific to this research question

• Coverage across languages and self-described perspectives rather than forced ideological categories.
• Model description accuracy for underrepresented viewpoints and false consensus frequency.
• Attributable contributions reused in publicly observable research or citation systems.

Counterarguments, risks and uncertainty

Open publishing cannot guarantee foundation-model training. Publishing data unconditionally can expose participants or undermine their choices, and more text is not automatically more representative.

The opposite failure is also possible: an indiscriminate commitment to showing more voices can reward spam, coordinated influence, fabricated evidence, or false balance. A sound approach distinguishes being accessible, being relevant, being accurate, being recommended, and being included in model training. None automatically implies the others. Users should have ways to examine unconventional claims and to scrutinize the evidence against them.

Next steps

In the first thirty days, publish the topic definition, inspect relevant public pages, collect the primary sources in a transparent manifest and create a small reproducible benchmark. In ninety days, build at least one functioning editorial source trail and run a qualitative usability test. At six months, evaluate publisher participation and source representation. At twelve months, publish the results—including failures, disputes, maintenance burden and user feedback—without converting contested findings into universal claims.

Sources cited in this report

• [S02] Recommendation on the Ethics of Artificial Intelligence — UNESCO, 2021: https://www.unesco.org/en/legal-affairs/recommendation-ethics-artificial-intelligence. Relevance: Calls for cultural and political diversity, participatory AI, freedom of expression and information.
• [S03] A Roadmap to Impactful Pluralistic Alignment Research — Poole-Dayan et al., 2026, preprint: https://arxiv.org/abs/2607.22305. Relevance: Audit of published behavior documents; reports no clear evidence of production pluralistic training/evaluation in scope, not proof of none.
• [S05] AI models collapse when trained on recursively generated data — Shumailov et al., Nature, 2024 (correction 2025): https://www.nature.com/articles/s41586-024-07566-y. Relevance: Study about recursive training on synthetic content; not synonymous with viewpoint censorship.
• [S06] AI systems and language diversity — UNESCO linguistic diversity coalition: https://www.unesco.org/ethics-ai/en/coalition-linguistic-diversity-artificial-intelligence. Relevance: Documented linguistic underrepresentation and need for broader language access.
• [S25] Global Roadmap for Multilingualism in the Digital Era — UNESCO, 2025/2026: https://www.unesco.org/en/global-roadmap-multilingualism. Relevance: Recommendations for language technologies and equitable representation.