Published on: 29/08/2024
My internship at IRC began with a deep dive into the integration of Artificial Intelligence (AI) in business processes and workflows. AI is a rapidly evolving and expanding field that has the potential to transform various sectors, including the development and WASH sectors. In this blog post, I share some of the main tasks I carried out during my internship at IRC, where I researched AI's integration in IRC's business processes and workflows, and developed custom chatbots.
Why AI matters for IRC
AI is a machine-based system designed to imitate intelligent human thinking and use data to make predictions and automate decision making. AI can offer substantial benefits for organisations like IRC, such as:
- enhancing productivity and quality of work by generating content, summarising documents, editing texts, translation, and creating meeting summaries. For example, Copilot for Microsoft 365 is a generative AI tool that offers an AI-powered Microsoft Office Suite that can help IRC staff with these tasks.
- enhancing, together with machine learning, decision making by analysing patterns of data and providing insights and recommendations. In the WASH sector, AI can help to forecast water demand and supply, monitor water quality and leaks. Beyond the WASH sector, AI can be used for disaster resilience and relief, disaster prevention and data-driven decision-making and governance in public affairs.
However, AI also poses some risks and challenges, such as data privacy and security, bias and misrepresentation, validation problems, and environmental data privacy and security, bias and misrepresentation, validation problems, and environmental impact. Therefore, it is important to implement and use AI responsibly and ethically, following guidelines and regulations (such as the new European Artificial Intelligence Act), and assessing the impact and suitability of each AI tool.
How I explored AI's potential for IRC
During my internship, I carried out several tasks to explore AI's potential for IRC, which include:
- Researching AI's role in the development and WASH sectors, and the current status and initiatives of AI at IRC.
- Surveys to assess the application, potential, and impact of Copilot for Microsoft 365, a generative AI tool that I piloted with six IRC staff members from different business units (see figs 1-3 below).
- Building three custom chatbots using the free version of Microsoft Copilot Studio, a tool that allows building conversational AI agents. The chatbots are: IRC Copilot, a general information chatbot that uses IRC's website as a data source; ICT Help Desk, a chatbot that provides IT support and guidance using SharePoint resources and external websites; and Policy Guide, a chatbot that helps IRC staff to access policy documents and information using SharePoint resources.
- Assessment of IRC's software tools to determine if they have AI capabilities incorporated, and identification of other AI tools that have potential for different business processes in IRC.
- Participation in online conferences and webinars on AI in the WASH and broader development sectors, where I learned from the experiences and insights of AI experts and practitioners.
- Interviewing AI experts in the WASH and other development sectors.
- Developing a business case for IRC management.
Fig. 1: Testers views on the use of Copilot
Fig. 2: Time savings of testers using Copilot
Fig. 3. Quality enhancement experienced by testers using Copilot
What I recommend
From the research, I argued that AI has great potential to improve IRC's work and impact. I concluded that IRC should strategically incorporate AI into its business processes to enhance productivity and workflow quality, and to keep up with the advancements and applications of AI in the WASH and development sectors. I also made the following specific recommendations:
- IRC should integrate Copilot for Microsoft 365, as it can save time and enhance quality of work for tasks such as content generation, document summarisation, editing, translation, and meeting summaries. The surveys conducted and usage scenarios I presented provided positive feedback and showed significant benefits from using Copilot.
- IRC should build custom chatbots, as they can facilitate efficient information and policy access for IRC staff, handle frequently asked questions, and enhance 24/7 accessibility and engagement with key sector audiences. The chatbots I have developed are examples of how chatbots can be used for internal and external purposes.
- IRC should engage with AI networks and working groups, as knowledge and experience sharing is crucial for implementing AI in IRC.
What I learned and gained from the internship
The internship was a forge where my skills were tempered and honed. AI was no longer a mere concept but a tool that I wielded with increasing finesse. Chatbot development emerged as a canvas for my creativity, while Power BI analysis became the lens through which I discerned patterns and insights from the data. The knowledge I amassed was not just theoretical; it was practical, applicable, and powerful. It was a transformative process that saw me evolve from an intern to a practitioner of AI.
The time I spent on the project far exceeded the official internship hours, but the return was immeasurable. I have emerged as an expert and equipped with the knowledge and experience to make a significant impact in the field. This internship was more than a learning experience; it was a transformative journey that has set the stage for my future endeavours in AI, data analysis and beyond.
Acknowledgements
I would like to thank the following people:
- Internship supervisor: Cor Dietvorst
- IRC Copilot testers and technical support: Sára Bori, Dechan Dalrymple, Laura Bosma, Danielle de Zwart, Esther Glotzenbach and Michael Abera (testers), Patrick van der Toolen (technical support)
- Experts consulted:Nick Dickinson (IRC Associate/WASHNote), Bénigne Mugwaneza Ishimwe (WASH Governance Expert, IRC), Olivier Mills (Baobab Tech/WASH AI), Ali Al Mokdad (Danish Refugee Council), Katy Sill (SEGURA Consulting LLC) and Daniela Weber (NetHope)
- Review and Editing: Cor Dietvorst and Tettje van Daalen