Education & youth IWIB4AI session Needs review

HackEthos 2025 with UCL – Innovating with Ethics: AI for Social Good

Students build AI prototypes for social and environmental challenges while learning to embed ethics into design.

Draft – legal and case law references not yet verified

Summary

UCL students and IWIB4AI’s UK Youth Chapter collaborated to develop ethical-AI prototypes addressing social and environmental problems. Mentoring supported ethical literacy and cross-disciplinary collaboration alongside technical development. The use case illustrates how practical AI education can encourage responsible innovation without treating a socially beneficial purpose as proof of ethical performance.

The Problem

Rapid prototyping can reward technical novelty before teams adequately examine affected communities, data practices or environmental costs. The challenge is to turn ethical awareness into demonstrable design choices and safeguards, rather than relying on an “AI for good” label.

Ethical Grounds

Fairness

Teams should examine who benefits from each prototype and who might be excluded or disadvantaged by its data, assumptions or outputs. A social-good objective does not eliminate the risk of unequal outcomes.

Inclusion

Cross-disciplinary collaboration can broaden problem framing, but it should not substitute for input from the communities a prototype aims to serve. Accessible design and opportunities for affected people to challenge assumptions should inform development.

Sustainability

Environmental prototypes should be assessed against their computing and resource costs as well as their intended benefits. Teams should consider whether a simpler, less resource-intensive approach could achieve the same goal.

Accountability

Mentoring should help teams document intended uses, limitations and responsibility for addressing harms. Prototype status should be explicit so that demonstrations are not mistaken for validated solutions.

Legal Compliance

Framework	Provision	Relevance
EU AI Act	Art. 2(8)	For an EU-facing development pathway, research, testing and development before market placement or putting into service are generally excluded from the Act, but this exclusion does not cover testing in real-world conditions.
EU AI Act	Art. 4	Where participating organisations qualify as providers or deployers within the Act’s scope, they must take measures to ensure sufficient AI literacy among relevant personnel, making context-sensitive ethical and technical mentoring pertinent.
GDPR	Arts. 5 and 25	Where EU GDPR applies and prototypes process personal data, teams must follow data-protection principles and implement protection by design and default; a UK-based educational activity alone does not establish EU territorial applicability.

Sources

Court & Regulator Decisions

IMY – Skellefteå municipality facial recognition in school (opens in a new tab)

Swedish Authority for Privacy Protection (IMY) · DI-2019-2221 · 2019

The unlawful school facial-recognition pilot shows why educational experimentation still requires privacy assessment and cannot rely on consent undermined by power imbalances.

Key Takeaways

Translate ethical mentoring into documented design requirements, tests and safeguards.
Involve intended beneficiaries and assess who could be excluded or harmed.
Use data-minimising approaches and compare AI with simpler, lower-impact alternatives.
Clearly label prototype limitations and reassess legal obligations before real-world testing or deployment.

From the Session