Danil Mikhailov, Executive Director of Data.org, on AI for Social Impact. Established five years ago by the Rockefeller Foundation and the Mastercard Center for Inclusive Growth, Data.org is a nonprofit dedicated to advancing the use of data and AI for social good. 

The rapid evolution of AI and data science presents both an unprecedented opportunity and a growing challenge for the social impact sector.

While AI-powered tools have the potential to enhance decision-making, streamline operations, and increase efficiency, the gap between the private sector’s adoption of AI and the ability of nonprofits to leverage these technologies remains significant.

One of the most immediate impacts of AI on data work is its ability to automate many traditionally labor-intensive tasks, from data cleaning and visualisation to sophisticated data analysis. For social impact organisations, this represents a powerful efficiency boost, particularly for those with limited resources. Yet, while AI can enhance accessibility to data and streamline its use, it cannot replace human judgment, particularly in contexts involving vulnerable communities. The ethical deployment of AI remains paramount, and organisations must ensure that human oversight is preserved in critical decision-making processes.

Beyond efficiency gains, AI is also reshaping how nonprofits and global grant-making organisations assess impact. Many NGOs possess vast repositories of historical data that remain largely untapped due to resource constraints. AI-driven document analysis and natural language processing are now unlocking these archives, enabling organisations to extract meaningful insights and make data-driven decisions. 

The conversation also delves into the broader ethical considerations of AI, particularly the risks associated with overcorrection in training data. AI models are designed to reflect the information they are fed, and any attempt to engineer ethical biases — whether to correct for historical exclusions or to impose specific viewpoints — must be handled with caution. The balance between mitigating bias and preserving accuracy remains a complex challenge, as evidenced by recent controversies over AI-generated historical imagery that distorted reality in the name of diversity.

The takeaway is that ethical AI cannot be an afterthought. It must be integrated into the design and development process from the outset, ensuring that social scientists, ethicists, and technologists collaborate in real-time rather than operating in silos.

Thank you for downloading this episode of the Do One Better Podcast. Visit our Knowledge Hub at Lidji.org for information on 300 case studies and interviews with remarkable leaders in philanthropy, sustainability and social entrepreneurship.