Seventeen nonprofit organizations participated in NetHope's European Chapter meeting in Woking, UK, at Plan International's headquarters, February 26-28, 2020.
Last week at the NetHope European Chapter meeting, 30+ representatives from 17 NGOs came together to share, learn, and collaborate. At this meeting, NGOs wanted to explore artificial intelligence in a hands-on, practical way and focus on an AI application with broad use in the social impact sector: chatbots, or software applications that use conversational AI (Natural Language Processing) to understand what a human wants and provide meaningful answers.
In the session, global NGOs had the opportunity to evaluate chatbots for the specific problem statements that their organizations are working on (e.g. health, education, emergency response, conservation) and very quickly get started with developing their own chatbots using a no-code tool from Microsoft. We also talked about risks and how to develop chatbots in an ethical, responsible, and sustainable way.
Here are some of the questions we explored:
The first question any NGO should ask itself is, are chatbots the right solution for the problem and conditions faced? Here are some of the needs that could be addressed by chatbots:
- Reach more people (e.g. refugees, youth) with services and information they need, such as educational content, legal information.
- Respond to repetitive information requests.
- Provide equal support to all people in need (e.g. youth in remote areas, girls).
- Support rapid learning about end-user needs and information gaps.
- Deliver high-quality, accurate information.
- Provide 24x7 coverage and reduce the time it takes to get answers to people in need.
- Support training needs of NGO staff.
What you should consider before getting started
Chatbots are only as good as the information they provide and processes that support their development, implementation, and maintenance. For NGOs, a whole set of factors are necessary to think about before getting started. The value and impact of chatbots on our work and the communities we support is ours to manage.
Here is a checklist of things you’ll need to consider to design your chatbot in a relevant and responsible way:
- Determine whether and why you need a chatbot. Start with the problem that needs to be solved and involve the end-user and impacted communities from the beginning.
- Learn from other chatbot implementations, reuse whenever possible.
- Develop chatbot persona and conversation flow early on.
- Decide what bot framework to use based on the scope and resources. Start small (i.e. content, features, number of users) and iterate before scaling.
- Understand potential security risks and develop a data plan—collection, policies (e.g. local privacy laws), security, and maintenance.
- Measure performance of your chatbot and iterate based on the end-user feedback.
- Embed your chatbot in the existing programming to ensure reach and sustainability.
- Ensure infrastructure for your chatbot to be accessible and useful (e.g. connectivity, power, devices).
- Have a transition plan from pilot to scaling, especially if the pilot was developed with/by a partner (e.g. a private sector company, an academic institution).
- Develop a maintenance plan to refresh content, fix technical issues, add new features, and secure resources for ongoing maintenance..
How do you get started without specialized technical expertise?
While bot technology has been around for a few years, there are several reasons for the increase in its development and use including significant advancements in Natural Language Processing (NLP). NLP is making it possible for bots to understand end-users’ needs and how to respond to them in a natural way. The evolution of tools has changed from requiring deep technical expertise to new tools that enable anyone in the organization (e.g. education program specialists) to create and maintain a chatbot as part of their program.
In the session at the European Chapter Meeting, Michael Tjalve, Principal Program Manager at Microsoft, provided a step-by-step demo of Microsoft’s new no-code tool: Power Virtual Agents. For NGOs to get started with AI, it’s important to have access to such no-code or low-code tools that don’t require specialized expertise.
What are some of the examples of chatbots in the international development sector?
[Source: NetHope members, IREX, and DIAL report ]
- TESSA: Helps marginalized youth in the Philippines articulate their skills, create a full competency profile, and find work and training opportunities where they live.
- Digital Care Assistant:Supports SOS Children’s Villages’ care staff around the world by answering their questions about pedagogical topics.
- MomConnect: Provides pregnant women and new mothers with a text-based help desk to receive tips & guidance throughout their pregnancy.
- Miss Migration: Provides migrants in Myanmar/Nepal with accurate information about the migrant rights and processes.
- LucyBot: Provides young women and men with information about sexual/reproductive rights.
- Farm.ink: Connects farmers to information and each other.
- Tarjimly: Provides refugees and humanitarian staff with instant access to live translators.
Chatbots are quickly becoming a popular method for organizations to interact with the communities they support and their staff. Whether used in humanitarian or development environments, or in the context of conservation, we invite organizations developing or using chatbots to share their experiences and lessons learned. This collaborative information-sharing will help advance responsible chatbot use. You can share with us by emailing firstname.lastname@example.org.
Special thanks to NetHope members including Thomas Rubatscher (SOS Children's Villages), Hyacinth Umaran (Plan International), and Amir Shiva (Norwegian Refugee Council) for their insights on the Why use chatbots and How to get started.