The Politics of AI Are Getting Contentious: How Nonprofits Can Separate Fact From Fiction & Build Trust

The politics of AI (artificial intelligence) are getting contentious, and nonprofits increasingly find themselves in the middle of debates that are growing acrimonious.

Just this week, popular YouTube science educator Hank Green admitted to using ChatGPT for video research, prompting a revolution of sorts among his fans, igniting a fierce debate on Reddit, X (formerly Twitter) and TikTok. There’s a lesson here for nonprofits, but many of the people on both sides of the debate reaching for pitchforks are missing the point.

There are two extremes developing on either side of the AI argument in social impact, and nonprofits are increasingly trapped between them. I will outline those extremes here:

The Pro-AI Absolutists Argue:

• AI will transform how we optimize and measure social impact in schools, cities, humanitarian contexts, and more.
• AI democratizes knowledge, learning, and skills, giving vulnerable communities access to compute prowess and expertise that seemed impossible just four years ago.
• AI saves time and money for nonprofits and NGOs that have neither.
• AI is inevitable, and nonprofits can’t afford to fall behind like they did with Web 2.0.
• Nonprofits can uniquely benefit from AI to reduce overhead and administrative costs while speeding up grantwriting and research and boosting communications and fundraising campaigns.

The Anti-AI Absolutists Argue:

• Anyone who uses generative AI is selling out to capitalism.
• Data centers should be the bipartisan political issue of our day, and we should be uniformly opposed to them.
• Every single word put into a generative AI tool implicates your organization in the climate and water crisis, and is incompatible with the mission of social impact.
• Generative AI displaces workers, artists, creators, and human knowledge itself.
• AI hallucinates, introduces bias, and has created an arms race between state and non-state actors who might use AI for malicious or violent ends.
• AI represents a privacy and surveillance nightmare for individuals, and its use by governments, both national and local, represents a threat to public freedom.

These debates are playing out inside nonprofits themselves, on public venues like LinkedIn, and are shaping politics during an important midterm election. More than ever, these debates are also generational. According to a recent Pew Research Center survey, younger Americans are nearly 10 points more likely to say that AI’s impact will be negative over the next 20 years.

I’m here to tell you that both sides have a point. I might even self-identify more with the pessimistic side of the aisle, mostly because our government’s track record of compellingly regulating new technology is, frankly, well…abysmal. But I also think that nonprofits simply cannot afford to sit on the sidelines while technology passes us by. We in the nonprofit community need to be active participants in shaping the future of what this technology looks like, both within our organizations and our communities. Put simply, it is not tenable to ignore this new technology, pretend it doesn’t exist, or bow to the crowds who act like mere experimentation with it is net evil. What this looks like is developing frameworks for ethical experimentation and thoughtful adoption internally, and creating a compelling and transparent case for (still cautious and ethical) AI use to public audiences and communities.

First, Here’s What I Think About The Debate:

• AI does have the ability to dramatically change nonprofit and NGO workflows for the better, but adopting new technology in a way that ensures it’s a net benefit (to time, workflows, finances, etc.) can be trickier than people tend to admit.
• Nothing about AI is inevitable except that models will get better. These models aren’t going backwards; they’re getting smarter, cheaper, and more efficient. If your argument against AI is “it doesn’t work yet” you’re eventually going to be proven wrong. But its scaffolding and ethical adoption are far less certain, and that is precisely where social impact organizations need to be operating. Ensuring that AI use internally and AI adoption externally are safe and beneficial is of urgent importance.
• AI will shape the future of work, and it is extremely dangerous that just a handful of companies have such a dominant role in dictating its adoption. Layoffs are likely inevitable, and that’s a horrible ramification. If your organization cares about AI governance, however, current debates around open source models (and whether they could be useful for your organization) are important. Nobody benefits from being inflexibly process-dependent on a single tool like Claude Cowork.
• AI does pose a threat to knowledge and to the very humanness of nonprofit organizations that misuse it. AI cannot replace human creativity if you believe that creative output requires an understanding of the human experience.
• The AI climate and water concerns are vastly overstated when framed from the position of individual responsibility. Articles claiming that policing the word counts of your AI prompts will minimize your impact on the environment are misleading. (Moderate daily AI usage really does not meaningfully change your energy burden as an individual consumer.)
• Concerns around AI data centers are real and valid. Widespread AI adoption can strain grids, impact local communities immensely, and create genuine environmental risks in the places where data centers are built. At an international level, AI infrastructure is the global economy’s hottest speculative investment, and it will have environmental ramifications. Importantly, data centers will eventually be built somewhere, however, and their off-shore construction may mean a whole host of new ramifications.
• The rights-based concerns about AI-enabled surveillance, AI targeting and autonomous weapons in conflict, general privacy, and AI’s nefarious capabilities (hacking, disinformation), as well as its implications for issues such as youth mental health, are extremely serious and all require urgent action and advocacy from nonprofits.
• Blanket bans on AI, like social media bans, will not solve the problems they purport to address. Moreover, communities that have historically been disadvantaged stand the most to lose from increasing tech inequity. Difficult problems are typically not solved by blanket or obvious policy fixes. Nonprofits should eschew advocating for outright tech bans, but instead engage the communities they serve to understand the concerns, risks, and opportunities emerging tech offers and foster dialogue.

It is no secret that Americans are increasingly souring on AI; poll after poll shows it. Fundamentally, however, there is a difference between a human rights nonprofit using AI to help with database management or build an anti-LGBTQ legislation tracker, versus a health nonprofit flooding the internet with AI-generated slop, or a humanitarian NGO inundating subscribers with biased AI-generated pictures of children in poverty.

There are good and bad AI uses. And good and bad reasons for using AI in the first place. It may be helpful to explain to your audience how you are using AI safely and responsibly to drive impact.

My Recommendation:

Write an internal AI policy. Internal AI-use policies should discuss things such as tools, guardrails, transparency, and safety, as well as involve employees in the process of shaping what responsible AI use and experimentation should look like. It should read like a detailed how-to guide, emphasize safety, and engender buy-in so staff feel comfortable asking questions, pushing back, and co-developing a process that feels inclusive. Learn more about writing great AI policies.

Write an external AI policy. This will be different from an internal policy. Your primary goal here is to build trust. Consider a hypothetical youth services organization:

• Emphasize your bright red lines: “We absolutely do not use AI in any moment where we are engaging with our youth community; we have strict security protocols in place to ensure privacy and safety for all participants within our youth organization.”
• Where is AI driving impact? “As we are a small organization, we use AI sparingly but efficiently to assist us with tasks such as spreadsheet management and data reporting within our administrative responsibilities. AI also supports our ability to identify and apply to grant opportunities, ultimately increasing our ability to secure funding for our youth programming.”
• How do you consider harms in communities you serve? “We acknowledge that in some contexts AI has the potential to harm our youth community, particularly as it relates to mental health. We have developed training to help staff reach out to children who we suspect, lacking other resources, may be turning to AI companions for help during difficult moments.”
• Build trust when you communicate. “All communications from us are really from us. We have spent years building a community based on mutual trust and respect, and part of that trust is pledging to always communicate with you in emails, phone calls, and other materials with the humanness you expect.”

I might be wrong, but I think the above approach is hard to argue with. Transparency goes a long way, and both the absolutist cynic and the absolutist optimist have something to appreciate about our hypothetical disclosure scenario.

Conclusion

The politics of AI are going to get messier before they get clearer, and nonprofits should not sit this debate out. Organizations have to be honest with staff and communities about where your red lines are, what your approach to AI is, and why human-based trust trumps all.

Find a vetted nonprofit consultant → nonprofit.ist