AI for Social Good: A Taxonomy for Supporting Employees in Non-Profit Organizations in the Office Sector

What is it about?
The potential of artificial intelligence (AI) to support office work in nonprofit organizations (NPOs) is still largely unexplored. This master's thesis develops and validates a practical taxonomy based on a socio-technical perspective for the systematic classification of AI functionalities relevant to office processes in NPOs. Based on literature and interviews, the taxonomy provides a structured framework for the strategic planning and implementation of AI in administrative tasks, helping NPOs to increase their efficiency and free up resources for their social mission.

The Research Contribution
NPOs are facing increasing administrative burdens with limited financial and human resources. AI offers promising opportunities for optimizing office processes, but practical guidelines for implementation are lacking. This thesis provides NPOs with a practical decision-making tool to evaluate and prioritize AI applications according to their needs and capacities. The research contributes to closing the gap between technological innovation and operational practice in the underrepresented NPO sector.

Key Findings
The result of the study is a validated taxonomy with eight dimensions: office processes, AI functionality, user acceptance, user interaction, technical integration effort, training effort, ethical integrity, and accessibility. Interviews with Blindspot - Inclusion and Diversity, a Swiss NPO, confirmed the relevance and applicability of the taxonomy. Specific use cases for AI were also identified, such as automated document processing, content creation, and fundraising support, which underscore the significant potential of AI to reduce workload and improve organizational effectiveness.

Practical Recommendations

• Use the taxonomy to evaluate and select AI tools systematically
• Identify priority use cases for administrative process optimization
• Incorporate accessibility and inclusion into AI solution design
• Address ethical aspects such as transparency and accountability
• Engage employees early to improve acceptance and reduce resistance

Methods
The thesis followed a mixed-method approach. First, a systematic literature review was conducted to identify relevant concepts and technologies. An iterative taxonomy development process, based on Nickerson et al. (2013), was applied, combining conceptual and empirical steps. Six qualitative interviews with employees of the Swiss NPO Blindspot validated and refined the taxonomy. Additionally, the interviews helped to identify specific AI use cases in the office context of NPOs. The taxonomy was enhanced with an accessibility dimension based on the findings.