You can view the webinar on demand and download presentation materials from our webinar library. Some elements of this article were developed using Microsoft’s Copilot AI tool.

While researchers have worked toward artificial intelligence (AI) for decades, the public release of highly capable “large language models,” such as ChatGPT, starting in 2022, kicked off the current whirlwind of development, investment, discussion, and debate. Prior to the release of these tools, for most people, AI was the stuff of computer science and science fiction. Now it seems to permeate everyday life, with some people believing it to be a potential savior of humanity, while others predict the opposite.

Stepping back, however, and looking at AI through the lens of potential social impact, we can identify both possibility and threat, and provide a path forward for nonprofit organizations looking to use AI tools in their work. If those organizations discuss AI opportunities openly, honestly, and inclusively, they can ensure that the technology serves the people and causes at the heart of their social missions.

AI Features and Trends: A New Wave of Tools

According to Jake Porway, co-founder of DataKind and Bridgespan’s AI fellow, three attributes define the latest wave of AI. Versus tools of only a few years ago, today’s AI tools are:

1. Generative: Where previous AI tools would make selections for users—such as the Netflix recommendation tool choosing a movie they might like based on previous choices—new AI tools can generate content based on user requests.
2. Conversational: Rather than requiring users to navigate menus or code to get the results they want from a piece of software, now they can have conversations in normal language to manipulate the tools. 
3. Scalable: Taking full advantage of previous AI tools usually required organizations to have their own AI or data science teams to build algorithms and create new software specific to the organizations’ needs. The latest tools, however, are general purpose and can be applied to many different tasks without needing a technology team—or even much technology knowledge. 

These new facets have opened a wide range of new possibilities for AI in the social sector. But AI doesn’t need to feel overwhelming. In fact, Porway says, AI is simply the next step in a long line of computing trends with which we’re already familiar. 

By recognizing this progression, taking lessons from how their organizations dealt with other technological advances, and adopting a framework to identify potential AI applications, organizations can strategically integrate the technology to enhance efficiency and effectiveness.

First Principles for Introducing AI

Before an organization dives into the possibilities for AI, Nabiha Syed, the executive director of Mozilla Foundation, advises that organizations ground their discussions in some guiding principles. 

1. Do No Harm - Bias: The potential for AI tools to do great good is balanced by the potential for great harm if they aren’t implemented thoughtfully and inclusively, Syed noted. In particular, organizations need to be aware of the potential for algorithmic bias in the tools they use—but this awareness doesn’t mean avoiding the tools entirely. “If you're an organization that collects data on children,” she says, “you're going to be in a totally different headspace when you're thinking about how to adopt and roll out innovations in [AI].” But, she says, if an organization isn’t dealing with similarly vulnerable populations, they have more latitude in how they utilize the tools.
2. Do No Harm - Inclusiveness:  Syed also noted that organizations need to be aware of issues around inclusiveness if they start implementing AI tools. In a world where the digital divide still exists, “you really shouldn't be building tech about people without them [involved].”
3. Intentionality: Syed added that organizations should be intentional in how they’re using AI. “You need clear change management processes,” she advises. “What's a problem for you? What's a bottleneck for you? What's an endless meeting that you're like, ‘this could have been an email or a process?’”  By starting with specific issues that AI might be able to solve, organizations can focus, build process, provide change management, and offer guardrails rather than dealing with “an amorphous AI cloud.” 

Frame Your Thinking about AI

Organizations trying to get their arms around AI’s potential can benefit from thinking about three categories of opportunities: 

1. Back office: This includes HR, communications, general “keeping the lights on” work. Generative AI tools can be useful here for such tasks as resume review, writing policy summaries, grant writing, translation, and content creation. “Writing and reading and summarizing, that's everyone's low hanging fruit right now,” according to Porway. “[Things like,] ‘help me summarize a legal document,’ or ‘help me write a contract.’” He also noted that AI will likely have some of its earliest and potentially largest impact on fundraising, with tools being created that will help answer questions like “are my donors about to stop contributing” and identifying those who are before they stop giving so an organization can try to keep them engaged. 
2. Program execution: This includes the actual methods by which you deliver on your theory of change. This includes opportunities such as digital service delivery through AI-powered chatbots, case management, staff training, data analysis of advocacy messages, member onboarding and recruitment, and more. 
3. Measurement, evaluation, and learning/strategy: How well are you delivering on your impact goals and how do you adjust your strategy to keep getting better?  Porway pointed to a specific example, Nonprofit AMA, that allows people interested in nonprofits to ask questions such as “How do I start a nonprofit in Des Moines?” Such tools, he says, aren’t going to replace a full strategy development process, but they could be useful for a nonprofit leader who’s thinking, “Hey, I can't get my board or my advisors together, and I just want to start some brainstorming or research.”

What’s Next in Your AI Journey?

Filling these three buckets with ideas may prove easy. But there will be challenges along the way. Change management is critical, as organizations must foster a culture of learning and adaptability to embrace AI. This involves investing in training and creating a safe space for experimentation. 

And, as Syed notes, performing adequate due diligence around data privacy and policy issues is critical when contemplating AI-based solutions. Beyond that, organizations should establish robust data governance practices and stay informed about legal and technological changes to responsibly harness AI's power.

Finally, the journey towards AI integration is not a solitary one. Collaboration with other organizations and stakeholders is vital. Nonprofits should seek partnerships that allow for knowledge sharing and joint problem-solving. Together, they can help drive the policies and best practices for using AI in the social sector and better tackle the challenges of potential bias and harm from AI use, ensuring that technology amplifies their collective impact.