EU AI ACT 2024 - Oerview

General Principles

• Risk Categorization
• Reference: Article 6 – Classification of AI systems.
• High-risk systems include those impacting education per Annex III.
• Transparency
• Reference: Article 52 – Transparency obligations for AI systems.
• Accountability
• Reference: Article 23 – Governance and accountability in high-risk systems.
• Human Oversight
• Reference: Article 14 – Human oversight requirements for AI systems.
• Ethical Framework
• Reference: Article 9 – Risk management system requirements.

Data Handling

• Data Privacy
• Reference: Article 10 – Quality of datasets, aligned with GDPR (General Data Protection Regulation).
• Bias Mitigation
• Reference: Article 10 – Avoiding biases in datasets.
• Data Security
• Reference: Article 15 – Cybersecurity requirements for AI systems.
• Data Transparency
• Reference: Article 13 – Documentation requirements for high-risk systems.
• Consent
• Reference: Article 52 – Consent and communication obligations for users.

AI-Driven Learning Tools

• Content Accuracy
• Reference: Article 10 – Dataset quality assurance.
• Personalization
• Reference: Article 14 – Aligning personalization with ethical oversight.
• Feedback Mechanism
• Reference: Article 54 – Reporting and feedback mechanisms for AI systems.
• Inclusivity
• Reference: Article 10 – Diverse datasets to ensure inclusivity.
• Cultural Sensitivity
• Reference: Article 9 – Risk mitigation strategies, including cultural sensitivity.

Grading and Assessments

• Fairness
• Reference: Article 7 – Prohibitions of certain AI practices (unfair grading systems).
• Explainability
• Reference: Article 14 – Human oversight to ensure explainability.
• Error Correction
• Reference: Article 56 – Error correction and liability mechanisms.
• No Sole Decision-Making
• Reference: Article 14 – Human oversight in decision-making processes.
• Accuracy Validation
• Reference: Article 10 – Continuous testing for dataset and model accuracy.

Student Engagement

• Interaction Design
• Reference: Article 14 – Ensuring systems are designed for responsible usage.
• Feedback Personalization
• Reference: Article 9 – Personalization with fairness considerations.
• Privacy in Chatbots
• Reference: Article 52 – Transparency in AI communication tools.
• Monitoring Usage
• Reference: Article 15 – Usage monitoring and cybersecurity protocols.
• Emotional AI Limitations
• Reference: Article 5 – Prohibition of harmful or manipulative AI systems.

Training for Educators

• AI Literacy
• Reference: Article 9 – Training and awareness programs for stakeholders.
• Bias Awareness
• Reference: Article 10 – Educator training on dataset biases.
• Decision Oversight
• Reference: Article 14 – Human decision-making training requirements.
• Ethical Use Training
• Reference: Article 9 – Awareness programs on ethical AI use.
• Policy Awareness
• Reference: Article 23 – Internal policies for AI governance in organizations.

Procurement and Deployment

• Supplier Compliance
• Reference: Article 24 – Supply chain management obligations.
• Documentation
• Reference: Article 13 – Comprehensive documentation of AI system operations.
• Risk Assessment
• Reference: Article 9 – Risk management plan for deployment.
• Third-Party Audits
• Reference: Article 20 – Conformity assessments by third parties.
• Regular Updates
• Reference: Article 13 – Documentation to reflect system updates.

Special Needs and Accessibility

• Accessibility Features
• Reference: Article 14 – Inclusion of human oversight for accessibility.
• Support for Disabilities
• Reference: Annex III – High-risk systems, including those designed for disabilities.
• Language Support
• Reference: Article 10 – Dataset diversity, including multilingual data.
• Customizable Interfaces
• Reference: Article 9 – Designing customizable features for inclusivity.
• Assistive AI Validation
• Reference: Article 20 – Validation and testing for assistive technologies.

Long-Term Impact

• Future-Proofing
• Reference: Article 13 – Regular updates and adaptable documentation.
• Sustainability
• Reference: Article 9 – Environmental considerations in AI usage.
• Cost-Benefit Analysis
• Reference: Article 20 – Economic assessment during conformity checks.
• Scalability
• Reference: Article 13 – Design requirements for scalability.
• Continuous Improvement
• Reference: Article 54 – Reporting and adapting AI systems.

Stakeholder Engagement

• Parental Involvement
• Reference: Article 52 – Transparency obligations to inform all stakeholders.
• Student Participation
• Reference: Article 54 – Mechanisms for user feedback.
• Community Outreach
• Reference: Article 23 – Institutional governance including community input.
• Cross-Institution Collaboration
• Reference: Article 20 – Shared practices through audits and testing.
• Regulatory Liaison
• Reference: Article 62 – Coordination with regulatory authorities.