UNESCO Recommendation on the Ethics of Artificial Intelligence — provisions explorer

AI in Education

Policy Area 'Education and Research', para 101 — provide adequate AI literacy education to the public

Editor rationale: A dedicated named Policy Area mandating public AI-literacy education at all levels (paras 101-110). Governs under the convention; medium (soft law). Reconciled implicit→governs (iter-451); blind panel 3/3 governs.

AI in Employment

Policy Area 'Economy and Labour', para 116 — Member States to assess and address AI's impact on labour markets

Editor rationale: A dedicated named Policy Area with operative 'Member States should assess and address' labour-market text. Governs under the catalog convention (explicit topic-specific operative provision); medium (non-binding soft law). Reconciled implicit→governs (iter-451); blind panel 3/3 governs.

AI in Healthcare

Policy Area 'Health and Social Well-being', para 121 — employ effective AI for health and the right to life

Editor rationale: A dedicated named Policy Area with operative health-AI duties (paras 121-123: safety/efficacy regulation, bias oversight). Governs under the convention; medium (soft law). Reconciled implicit→governs (iter-451); blind panel 3/3 governs.

Transparency Obligations

Principle 'Transparency and explainability', para 38 — people informed of AI-based decisions + right to request explanation

Editor rationale: A named core Principle with operative transparency duties (para 37 ties it to human rights; para 38 grants the right to be informed + request explanation + seek review). Governs under the convention (cf. OECD transparency=governs); medium (soft law). Reconciled implicit→governs (iter-451); blind panel 3/3 governs.

Individual Redress

Policy Area 'Ethical governance and stewardship', para 55 — harms through AI investigated and redressed via enforcement + remedial actions

Editor rationale: Operative redress duty (para 55) reinforced by the Responsibility & accountability principle (paras 42-43) + the right to contest/correct (para 38). Governs under the convention (cf. OECD redress=governs); medium (soft law). Reconciled implicit→governs (iter-451); blind panel 3/3 governs.

Training-Data Rights

Policy Area 'Data Policy', para 71 — data-governance strategies ensuring continual evaluation of training-data quality

Editor rationale: A dedicated named Policy Area whose para 71 explicitly governs training-data quality/collection/security; paras 72-73 add privacy safeguards + data-subject rights. Governs under the convention; medium (soft law). Reconciled implicit→governs (iter-451); blind panel 3/3 governs.

Development-Rights Framings

Policy Area 'Development and International Cooperation', para 79 (+ Diversity Principle para 67) — AI-for-development bound to the values/principles

Editor rationale: A dedicated named Policy Area framing AI-for-development as rights/values-bound (para 79), esp. for LMICs/LDCs (paras 80-81), with the Diversity & inclusiveness Principle (para 67). Governs under the convention (cf. UN-Res development_rights_framing=governs); medium (soft law). Reconciled implicit→governs (iter-451); blind panel 3/3 governs.

International Coordination

Policy Area 'Development and International Cooperation', para 80 — platforms for international cooperation on AI

Editor rationale: A dedicated named Policy Area with operative international-cooperation duties (para 80; para 78 AI ethics in international fora). Governs under the convention (cf. Bletchley international_coordination=governs); medium (soft law). Reconciled implicit→governs (iter-451); blind panel 3/3 governs.

Environmental Impact of AI Training

Policy Area 'Environment and Ecosystems', para 84 — assess direct/indirect environmental impact incl. carbon footprint + energy consumption

Editor rationale: A dedicated named Policy Area + Sustainability principle with operative carbon/energy assessment duties (para 84) and an efficiency mandate (para 86) — directly reaching training's energy cost. Governs under the convention; medium (soft law). Reconciled implicit→governs (iter-451); blind panel 3/3 governs.

Biometric Identification

Proportionality & do-no-harm principle (AI should not be used for mass surveillance/social scoring) + Right to privacy principle (para 74, biometric data) — no dedicated biometric-ID provision

Editor rationale: Reached only via the general proportionality principle (no mass surveillance/social scoring) + data-sensitivity, not a dedicated biometric operative section. Kept implicit on reconciliation (the blind panel split 2-1 governs/implicit); the conservative call stands. Not 'conflicts': its surveillance-skeptic stance aligns with a biometric-ID-restriction framing.

AI in Criminal Justice

Ethical-governance section, paras 62-63 — names law enforcement + the judiciary as sensitive use cases requiring oversight; no dedicated criminal-justice regime

Editor rationale: Reached via the general sensitive-use-case + human-oversight framing (paras 62-63 name law enforcement/judiciary), not a dedicated criminal-justice Policy Area with topic-specific rules. Kept implicit on reconciliation; the conservative residual the blind panel (which read governs) flags.

AI-Driven Worker Displacement

Policy Area 'Economy and Labour', para 118 — fair transition (upskilling/reskilling) for at-risk workers; a sub-provision of the labour area

Editor rationale: Worker-displacement is a sub-provision (para 118 reskilling/transition) WITHIN the 'Economy and Labour' Policy Area already scored governs via 'employment' — not its own dedicated section. Kept implicit on reconciliation to avoid double-counting the area; the conservative residual the blind panel (which read governs) flags.