By Leila Toplic, Head of Emerging Technologies Initiative, NetHope, and Michael Tjalve, Principal Nonprofit Architect, Microsoft Tech for Social Impact
Since 2017, NetHope has focused on growing the capacity of nonprofits to evaluate, develop, and implement AI/ML and data in their work. Much of this work has been done in collaboration with partners like Microsoft.
Partnerships are key to timely, impactful, and responsible adoption of AI in the nonprofit sector. Partnerships with the private sector, academia, and others can unlock access to resources, knowledge, and a greater impact at scale. Building around collective capabilities, resources, and points of leverage, and working towards a common goal, partnerships create value. For example, partnerships can help nonprofits address problems faster through timely access to knowledge, technical expertise, and resources. In turn, partnerships with nonprofits can help technologists better understand real-world needs and use cases, and build products that benefit all.
In summer 2020, NetHope and Microsoft’s Tech for Social Impact team partnered to enable NetHope’s Africa Chapter NGOs to integrate conversational AI into their processes to help address some of the challenges that were exacerbated by the Covid-19 pandemic. As part of this project, several NGOs came together to develop chatbot solutions which will enable them to improve the way they provide services and support to their staff. To learn more about these projects, please read this blog post.
Here, we discuss some of the key considerations and practices for building responsible AI partnerships based on the Africa Chapter chatbots and past projects – and while this blog is focused on developing solutions using AI tools, the four considerations are applicable to using any new technology in the nonprofit sector.
1. Build partnerships close to the point of action
The best outcomes happen when projects are driven and owned locally. Why? Because it’s the local experts who hold insight into which needs require the most immediate attention, who understand which interaction model will ensure adoption by the end users, and who know which external factors need to be taken into consideration for successful project outcome.
We learned this on another project, where the chatbot development was led by external technical experts (due to lack of technical expertise locally) and their limited understanding of the context led to wrong assumptions in the program design (eg, hardware requirements), impacting the usability and adoption of the solution.
To make it possible for design, development, and implementation to happen at the point of action (ie, locally), it’s important to create an adequately enabling environment where conditions are optimized for successful outcomes. This takes us to the next consideration.
2. Invest in capacity building and provide adequate resources
A key obstacle to adoption of AI in the nonprofit sector is first knowing how to get started without specialized technical expertise and then learning how to sustain the solutions beyond the initial pilot. We saw firsthand with the NetHope Africa Chapter NGOs how an intentional and adequate investment in capacity building can enable problem-owners to become problem-solvers, learn how to sustain the solutions, and become trainers themselves (ie, cultivate local knowledge through sharing).
By being intentional about the design of the partnership to ‘fit’ the project needs, we were able to bring together the right combination of experts and stakeholders into the (virtual, as of late) room and ensure they had adequate support to succeed.
We incorporated capacity building throughout the whole project lifecycle - from the initial training at the July 2020 NetHope Africa Conference to hands-on, project-based skilling in everything from how to develop a robust user journey to how to use Power Virtual Agents tool to build a chatbot. We used a guide-on-the-side model, with NetHope facilitating and guiding the overall project and Microsoft actively engaged throughout the process to provide technical guidance and help resolve technical issues.
While the new low-code and no-code tools, like Microsoft’s Power Platform, open up opportunities for everyone to get started without having specialized skills — it’s the capacity building that unlocks the lasting impact of such tools by enabling nonprofit practitioners to develop the know-how to be self-sufficient in using the technology in an impactful and responsible way. They, in turn, can then train their peers on the technologies used which enables scale of know-how in a relevant (ie, NGO-specific) way.
In addition to capacity building, another important enabler of timely adoption of AI in the nonprofit sector is funding. Funding for getting started (ie, for staff time, product usage, infrastructure) is critical for new, unproven projects that leverage emerging tools and technologies.
We’ve found that when given the access to knowledge and funding, along with expert support throughout the project, nonprofits can build the solutions and capacity at the same time.
3. Design responsibly for a human-centered experience
While technology is a powerful tool and there is a lot of excitement about using AI in nonprofit programs, it should never be the starting point or the main focus for creating solutions to societal challenges. We believe that it’s important to focus on creating a whole solution that meets end-users where they are (eg, with onboarding materials, incentives, awareness campaigns, etc) and to carefully evaluate suitability of AI tools for the problem/context.
For the Africa Chapter chatbots project, we used an agile, iterative model that enabled us to learn and incorporate end-user feedback throughout the development process. We were also intentional about building responsibly from the start by incorporating ethical reviews and discussions throughout the project which allowed us to anticipate where things can go wrong and design solutions for the most optimal benefit of the end-users. For specific examples, see this blog post.
4. Drive for impact beyond a single project through reuse, sharing, and collaboration
In the Africa Chapter chatbots project, a key criterion for project selection was that it would be designed to be a templatizable solution, meaning that once this solution is released, another nonprofit organization can adapt the template for their own use (ie, their end-users, geographies). Templatization and re-use accelerate time to impact and avoid unnecessary and costly duplication.
In addition to the development of templates that can be reused and adapted for scale, we encouraged ideation and feedback on future technology product roadmap from the perspective of NGOs and the communities they serve.
These are just some of the characteristics of impactful and responsible partnerships for AI in the nonprofit sector, and we are already thinking about the ways to improve and build on what we know. We plan to continue to learn and iterate on this partnership model in future projects. One thing we know for certain, intentionality is key in developing effective partnerships. Partnerships that focus on building local capacity and supporting local teams can broaden the reach, accelerate impact, and ensure sustainability of programs and solutions. And we can do more, more effectively, and with less duplication when we work together.
For more information about nonprofit offers from Microsoft, please visit www.microsoft.com/nonprofit