Artificial intelligence in education policy: responsibilities across jurisdictions

Current consideration of artificial intelligence in education policy is informed by the Beijing Consensus adopted in May 2019, with consequences for governance, evidence and the treatment of affected learners. A policy instrument has practical effect only when its scope, responsible actors and relationship with existing law are understood. The unit of review should correspond to the full reach of the decision, including significant differences in provision and population.

In the context of artificial intelligence in education policy, the Beijing Consensus on Artificial Intelligence and Education was adopted in May 2019. It addresses policy planning, management, teaching, learning, skills, lifelong learning, inclusion, gender equality, data and research. It promotes human-centred and equitable use rather than technology adoption as an end in itself. Authorities and providers should therefore connect each proposed use to an educational purpose, governance responsibility and evidence of benefit and risk.

A proper review of implementation should establish the intended outcome before selecting controls or indicators. For the policy position, implementation should be assessed against observable effects on access, learning, safety and fair treatment, rather than against the existence of a policy statement alone. The record for artificial intelligence in education policy should explain why the approach suits the affected context, how material departures are authorised and when review will occur.

Policy context for artificial intelligence in education policy

The formal status of the Beijing Consensus adopted in May 2019 should be preserved in any public account. For artificial intelligence in education policy, the instrument should be used to identify the intended direction, the actors addressed and the implementation measures that remain necessary.

In reviewing artificial intelligence in education policy, technology may support teaching, administration and access, but consequential educational decisions must remain accountable, explainable and open to effective review.

• Review incidents and supplier changes, identifying the accountable function and affected scope.
• Test performance across relevant groups, identifying the accountable function and affected scope.
• Classify uses by effect on learners.
• Notify users of material limitations before using it to determine a learner or provider outcome.
• Control personal and confidential information.

Responsibilities and affected parties

The analysis of artificial intelligence in education policy should make its decision rule explicit. In this case, cross-jurisdiction interpretation should distinguish international commitment, regional instrument, national law, regulatory direction and provider policy. Each has a different source of authority and may apply to a different object or person. This supports consistent review and reduces the risk of redefining the basis of judgement after an adverse result appears.

A narrow control over the measure may create false assurance. In the present context, unequal performance across learner groups, unverified outputs entering teaching or assessment and loss of meaningful human review may produce acceptable aggregate reporting while individual learners remain exposed to material disadvantage.

In work concerning artificial intelligence in education policy, the evidential record should be limited to material that can answer the question under review. For the arrangements, the most relevant material is likely to include supplier change and incident records, data provenance and access controls, an inventory of systems and their intended uses, and records of human review and overrides. Each source has limitations; confidence depends on corroboration between independent records and transparent treatment of uncertainty.

• Do partner arrangements change responsibility?
• What is the status of the relevant instrument?
• Which jurisdiction governs the activity?
• Who has enforcement authority?
• How will conflicting requirements be managed?

Implementation risks

A competent review of the policy position should prepare a jurisdictional register identifying the service, learner location, provider location, responsible authority, applicable instrument and conflict rule. In reviewing artificial intelligence in education policy, obtain competent interpretation where the legal position is uncertain and do not resolve uncertainty through promotional wording. Within the scope under review, adverse cases and unresolved contradictions should be retained because they may reveal limitations concealed by an average result.

For artificial intelligence in education policy, a policy conclusion on the issue should state who is required or expected to act, the source of that expectation and the consequence of non-implementation. Any conclusion should state where differences in law limit its application. Communications should preserve the legal status and effective date of each expectation described.

Findings on the policy position should preserve material uncertainty and limits on application. In this case, a technical capability is not evidence that a use is educationally justified. For artificial intelligence in education policy, accuracy measured in one setting may not transfer to another population, language, curriculum or decision context. The existence of an international commitment does not remove the need for jurisdiction-specific interpretation, consultation and proportionate transition arrangements. A finding should not be separated from limitations capable of changing how it is understood or applied.

For decisions concerning artificial intelligence in education policy, decisions concerning the arrangements should remain traceable to the information available for the stated reference period. Changes in condition, evidence, method and interpretation should be recorded separately when a conclusion is revised.

Accountability for artificial intelligence in education policy should follow decision-making authority. Within the scope under review, operational tasks may be delegated, but accountability for material effects on learners must remain identifiable.

Data used for the policy position should be interpreted against stable definitions and an identifiable population. As regards artificial intelligence in education policy, reporting should identify a break in comparability before describing movement over time. Public confidence cannot be separated from an institution's ability to identify responsibility and substantiate its conclusions.