Co-Designing Islamic AI Ethics: Insights from the UK Muslim Community

Paper Title

Co-Designing Islamic AI Ethics: Insights from the UK Muslim Community

Publication Info

• Topic area: Ethical AI governance informed by Islamic values
• Keywords: AI ethics, Islamic values, co-design, UK AI White Paper, epistemic pluralism, decolonial AI, transparency, accountability, inclusivity, stewardship

Background and Problem

• Problem / challenge: Current AI governance frameworks are predominantly shaped by Western, secular values, leading to epistemic exclusion of non-Western perspectives, including Islamic ethical principles.
• Significance: Islam is the world’s second-largest religion with 1.9 billion adherents, yet Islamic perspectives are underrepresented in AI ethics, which risks marginalising Muslim communities and their values in AI development and governance.
• Motivation and related work: Prior work has highlighted the dominance of Western epistemologies in AI ethics, with limited inclusion of religious or non-Western perspectives. While some frameworks attempt to incorporate diverse values, they often fail to align with non-Western traditions. This paper addresses this gap by exploring Islamic values through co-design workshops with Muslim women in the UK.

Solution

• Proposed approach: Co-design workshops with 12 Muslim women in the UK to identify Islamic ethical values for AI systems and compare them to the UK AI White Paper.
• Novelty:
  1. Identification of four themes of Islamic ethical values for AI: Honesty, Transparency, and Trustworthiness; Knowledge, Responsibility, and Divine Accountability; Justice, Power, and Equity; Unity, Inclusivity, and Diversity.
  2. Introduction of Islamic processes such as Hadith authentication, consultation (shūrā), and wealth distribution (zakat) as actionable methods for ethical AI development.
  3. Empirical evidence from practising Muslim women to ground theoretical discussions of Islamic AI ethics.
• Procedure and key techniques:
  • Conducted two in-person workshops lasting two hours each.
  • Participants identified Islamic values relevant to AI ethics, mapped them to AI contexts, and critiqued the UK AI White Paper.
  • Used thematic analysis to extract key values and processes grounded in Islamic ethical reasoning.

Results

• Concrete findings:
  • Participants viewed honesty as foundational to transparency and trustworthiness, emphasizing the need for verifiable AI outputs and ethical data practices.
  • Islamic values such as knowledge-seeking, critical thinking, divine accountability, and stewardship were identified as essential for ethical AI.
  • Justice and equity were framed as addressing structural inequalities and environmental stewardship, extending beyond individual rights.
  • Inclusivity and unity were emphasized, advocating for plural perspectives in AI development and governance.
• Advantage over baselines:
  • Extends existing AI ethics frameworks by introducing Islamic values and processes that challenge Western assumptions and provide alternative approaches to accountability, fairness, and inclusivity.
• Experiments / evaluation:
  • Workshops with 12 Muslim women, using structured activities to identify values, map them to AI contexts, and critique the UK AI White Paper.
  • Participants rated their AI knowledge between 1 and 7, ensuring diverse perspectives from non-experts.
• Limitations and future work:
  • Gender-specific workshops may not capture the full spectrum of Islamic perspectives.
  • Findings are specific to the UK AI White Paper and may not directly translate to other frameworks.
  • Future research should explore diverse Islamic communities and contexts to validate and expand the proposed values and processes.

Summary

This paper identifies Islamic ethical values and processes for AI governance through co-design workshops with Muslim women in the UK. Key values include honesty, divine accountability, justice, and inclusivity, which challenge and extend Western-centric frameworks like the UK AI White Paper. Islamic processes such as Hadith authentication, consultation (shūrā), and wealth distribution (zakat) are proposed as actionable methods for achieving ethical AI. The findings contribute to calls for epistemic pluralism and decolonial approaches in AI ethics, offering a foundation for integrating Islamic perspectives into global AI governance.