Final Declaration of the Divan on Ethics of Artificial Intelligence

This document is the final declaration of the Divan on Ethics of Artificial Intelligence, conducted as an online Zoom meeting with 477 participants on June 22-23, from 1 to 5:00 p.m. EDT (19:00-23:00 CET). This declaration is not intended to be the final word on AI ethics, but instead aims to explore the issue and encourage action, deliberations, publications, and research on the subject.

Concepts

• Hizmet Movement: A civil society movement inspired by Islamic and universal human values, aiming to promote co-existence and service to humanity through education, dialogue, and humanitarian activities within the framework of social responsibility and philanthropy. It is based on voluntarism and embraces religious, social, and cultural diversity. (Source: AfSV)
• Hizmet members: Anyone who feels loyal to the Hizmet Movement and identifies with its values. For other fundamental concepts, refer to the Guide to Artificial Intelligence.

Refer to the AI Guide (Turkish) for others.

General Principles

It is recommended that the actions of Hizmet members (as it pertains to the concept of artificial intelligence) be shaped based on the following general principles:

1. Conscious approach: Like any transformative technology, the impact of AI depends on how it is applied. Legitimate concerns about AI emphasize the importance of audits, oversight, and ethical rules. Systems should be monitored, policies should be reformed, and individuals should act consciously and intentionally.
2. Jurisprudential measures: Islamic legal principles like kıyas (analogy), icma (consensus), istihsan (preference), sedd-i zerai (blocking means), and maslahat (public interest) should be utilized to outline AI frameworks. The protection of religion, life, property, intellect, lineage, and freedom are essential interests (maslahat). Therefore, the legal status of AI should be defined, and liability should be assigned to AI systems to protect these interests.
3. Status of artificial intelligence, removed.
4. Protection of the human person: The safeguarding of human dignity and freedom must be unequivocally prioritized during the development and implementation of artificial intelligence systems. In this process, it is essential that existing discrimination based on race, religion, language, gender or social class is not perpetuated, and new discrimination is not created.
5. Universal principles: Today, artificial intelligence technology and its legal framework is a matter of global concern. Universal values must be upheld when developing solutions. So, universal principles should be taken into account by both the AI developers and users of artificial intelligence. However, frameworks based on the principles of morality in the fundamental sources of Islam will have more sanctioning power and meaning for Muslims.
6. Structuring rules and laws: Lawmakers should create flexible and adaptable frameworks that define desired outcomes rather than detailing how to achieve them. Rules should be built on existing "hard and soft law foundations." The context of AI applications is important, and a risk-based approach that holistically assesses risks and benefits should be adopted.
7. National and global cooperation: Coordination and cooperation mechanisms should be established at national and international levels amongst lawmakers, and updated regulatory and oversight systems should be created based on these mechanisms.
8. AI Train: When creating rules and systems for AI in various fields, it is important to keep in mind that AI is evolving rapidly compared to these systems. To keep up with this pace, collaboratively developed and continuously revised regulations should first be tested in small-scale isolated environments and tested with different parameters until they become reliable.
9. Data: The data on which AI models are built is fundamentally important. Responsibility for this data extends beyond governments, lawmakers, companies, and technology experts to each individual. Each person should make conscious decisions about the data they share, understand its potential uses, and advocate for ethical practices in data processing.
10. Active participation in data production: Each individual should recognize their role in shaping a balanced AI ecosystem and should act ethically in creating digital content and respecting intellectual property rights during data compilation and usage. Contributing to digital content such as digitizing books, writing blog posts, contributing to open sources like Wikipedia, or producing videos should reflect our perspectives and values.
11. Transparency and accountability: AI applications should clearly and transparently disclose their decision-making mechanisms and operating principles to intervene in potential injustices and build trust. The internal workings of systems should be explainable to monitor and hold individuals or organizations accountable for damages caused by AI systems.
12. Robot law: It is unacceptable for artificial intelligence to make life-or-death decisions. Developed AI methods must be proportionate to legitimate purposes. The protection of living beings is a fundamental principle.
13. Environment: The high energy consumption and environmental impact of training AI models should be discussed, and appropriate frameworks should be drawn with attention to sustainability.
14. Human-AI collaboration: Hybrid human-AI teams, combining human intuition with AI's data processing capabilities, are more effective and reliable than individuals working alone. Human oversight can provide a safety measure against potential AI errors.
15. Privacy and trust:
    1. Personal data provided to AI models should be protected by mechanisms that allow access only to authorized persons.
    2. The appropriate technical and organizational measures should be taken for data protection, and only necessary personal data should be processed.
    3. No AI model should be used as a tool to suppress or eliminate fundamental universal and human rights.
    4. Individuals should be informed about how their data is processed and should have the right to stop and delete its use at any time, even if they have previously given consent.

The following items have been placed under headings taken from AfSV’s “Core Values of the Hizmet Movement.” This section outlines the various ways that Hizmet members could benefit from artificial intelligence technologies.

Education

16. Education and learning: AI-supported applications can provide personalized educational materials based on users' learning styles.
17. Language learning: Artificial intelligence-based language learning applications can personalize grammar exercises according to the user’s level and provide feedback by analyzing their pronunciation.

Spiritual Nutrition and Development Activities

18. Counseling and guidance: AI can be useful in counseling and guidance services. AI systems can be developed to provide specialized guidance and counseling tailored to each individual's needs. For example, it could be useful to develop chatbots that provide responses that are founded in Hizmet principles.

Dialogue (Social Peace and Harmony)

19. Data purification: Since machine learning can infer malicious judgments beyond human understanding, AI systems must be free of discriminatory, inaccurate, and low-quality data.
20. National standards: A national standards institution should be established to create mechanisms for licensing and certifying new AI models and safeguards to protect human determinism.
21. Joint projects: What other religious/ethnic/social groups are doing in AI should be closely monitored, and joint projects should be carried out together when necessary.

Humanitarian Aid, Social Welfare

22. Equal opportunity: The continuity of social welfare is closely related to the safety and reliability of AI systems. Success and professional opportunities for human welfare should be provided for everyone, not just elites.
23. Positive transformation: Existing social and economic inequalities should not be perpetuated when developing AI systems. Changes in traditional job markets must not disturb social balance. The transformation should prioritize human welfare and fair progress as a fundamental principle.
24. Impacts: AI should not negatively impact justice, freedom, equality, or democracy. The environmental impacts of intelligent systems should be evaluated, and long-term benefits should be targeted.
25. Needs analysis and resource allocation: Hizmet institutions should analyze collected data through AI to determine which regions need which types of aid, ensuring more effective resource allocation.

Sports, Culture, Art

26. Health and sports: AI-supported health applications can personalize exercise routines, health screenings, fasting, and diets and can adjust them according to religious considerations.
27. Culture and art: AI can help people participate more productively in their communities by suggesting activities based on their interests, needs, and culture. AI can showcase art and historical artifacts from different cultures online, allowing people to learn about different cultures.

Publications

28. Content production: AI can automatically generate content in standard formats, such as news articles, sports reports, or financial summaries. It can also create social media posts or advertising texts.
29. Committee: A committee can develop a regularly updated guide on AI ethics. This committee should consist of individuals knowledgeable in both Islamic and AI-related matters. The guide should offer solutions for different situations and should be shared with the public.

Presented to the public.