Last week, Microsoft’s Tech for Social Impact Team, LMU Munich, and our team convened at Microsoft’s Munich headquarters to discuss advancing Responsible AI. This meeting highlighted a shared commitment to ethical, interdisciplinary approaches to artificial intelligence, focusing on its potential and societal implications.

 

Establishing the Research and Development Group

A key outcome of the meeting was the formation of the Research and Development Group, an interdisciplinary group aimed at addressing the ethical and practical aspects of generative AI. The discussions focused on aligning goals, refining objectives, and planning for future developments.

Activities during the meeting included:

​

• Strategic Project Planning: Defining objectives for advancing Responsible AI.

• AI Architecture Scoping Workshop: Outlining the technological structure of AI systems.

• Framework Development Discussions: Creating a guiding structure for initiatives.

• Empirical Research: The group initiated plans for empirical research in partnership with LMU Munich.

 

Developing a Framework for Responsible AI

The Research and Development Group is building a framework for Responsible AI, guided by three key perspectives:

1. Legitimacy: Rooted in a human rights-based normative foundation.

2. Legality: Complying with the EU AI Act and other regulations.

3. Domain-Specific Value: Leveraging insights from fields such as education to enhance AI's impact.

 

The framework will be developed iteratively, starting with a sustainability feature that embeds environmental considerations into AI systems. Each technological feature will be refined based on practical insights, ensuring adaptability and effectiveness. Using Azure AI as the technological foundation, the framework will balance normative and technical aspects. Collaboration with Microsoft’s Tech For Social Impact Team will strengthen the integration of these principles into Azure AI tools.

 

Research and Participatory Design

The Group’s framework development is grounded in practical applications, supported by empirical research conducted with LMU Munich. These studies examine AI's social dimensions, metadata, and user impacts, ensuring the framework aligns with real-world needs. An iterative design approach is central to the Group’s methodology. Insights from early research and feature development, such as the sustainability feature, will guide subsequent iterations. This ensures continuous improvement while maintaining scalability and applicability across sectors.

 

Open-Source Development and Future Directions

To achieve its goals, the Research and Development Group is expanding its team and prioritizing open-source innovation. Supported by ongoing scientific research and publications, the initiative aims to develop tools and frameworks that address ethical challenges and practical requirements.