AI Literacy for Social Impact with Vilas Dhar
In this episode of All Tech Is Human’s This Month in Responsible AI, Rebekah Tweed sits down with Vilas Dhar, President of the Patrick J. McGovern Foundation (PJMF). The conversation explores the role of AI literacy, pathways for responsible AI adoption, the function of philanthropy in shaping AI governance, and how nonprofits can leverage AI to drive systemic social change.
All Tech Is Human recently released five short Responsible AI courses that you can take for free here.
Key Takeaways from the Livestream
1. AI Literacy Is About Power, Not Technical Knowledge
Vilas reframes AI literacy as civic empowerment: it should help communities understand how AI influences political and social decision-making, ensuring that everyone (not just technologists) can shape the future. This has been a longstanding position of All Tech Is Human, which has argued that we need a vibrant multidisciplinary Responsible Tech ecosystem in order to tackle complex issues.
“AI literacy is not a technical skill set. It’s a fundamental conviction about a democratic process.”
“We rarely start from learning about lines of code. For me, this is a literacy that leads to confidence and empowerment that lets people shape decisions about how AI will shape our lives.”
2. Society Needs “Institutional Imagination”
We are at an inflection point where technical literacy is not enough. New governance frameworks, norms, and democratic structures must be built to avoid repeating past harms from earlier technological revolutions.
“We have to move very quickly to a new form of AI literacy that goes from how these tools work to how we are going to govern them.”
“Every major technological transformation requires new institutional frameworks. What do we need to put in place today to ensure we are safe and secure tomorrow?”
“Too often, the story of AI is told through hype or horror. What’s missing is hope—hope in human institutions that could use these tools to do good.”
“We’re still not thinking big enough as a sector. We have to expand our shared thinking about what's possible.”
3. Nonprofits Have a Unique Role in Shaping Responsible AI
Nonprofits can use AI not just for operational efficiency but to reimagine solutions to entrenched social problems. We can start shifting from scarcity thinking to systemic redesign. Vilas challenges the idea that AI is just a productivity tool or efficiency booster for nonprofits. Instead, he sees nonprofits as uniquely positioned to ask the larger systemic questions others often ignore. These raise questions about:
Dignity
Equity
Scarcity
Structural injustice
AI literacy, in this context, isn’t about just learning to use tools. It’s about learning how to design more just systems, and how nonprofits can leverage AI not simply to work faster, but to redefine what is possible. Vilas argues nonprofits should feel hope rather than hype or horror: hope rooted in human capability to use new technologies for societal good.
In a similar vein, All Tech Is Human views technology as neither a panacea nor a tragedy; as our name suggests, technology is impacted by humans while also impacting humans. Therefore, we fall under “tech pragmatism,” where it behooves us to come together as a society in creating norms, guardrails, and education.
“Nonprofits don’t just exist to fix market failures. They can ask: why does the world have to be this way in the first place?”
“The real opportunity isn’t to ask how our organization uses an AI tool more efficiently. It’s to ask: if we could use these tools, what would we build that fundamentally changes the problem we’re trying to solve?”
4. The Nonprofit AI Journey Has Five Stages
Vilas outlines a practical framework nonprofits can use:
Data and organizational readiness
Translating mission into use cases
Tactical planning
Governance and accountability
Implementation with real-world feedback loops
5. Common Pitfalls Include Data Gaps, Siloing, and Limited Ambition
Nonprofits often underestimate data challenges, treat AI as a tech-only issue, and focus on small efficiency improvements instead of systemic transformation.
Across nonprofits, Vilas sees four repeated challenges:
Belief that AI “isn’t for us,” or delegating it solely to tech staff rather than linking it to mission and strategy.
Underestimating data readiness, leading to shaky foundations for future AI work.
Delaying governance conversations, instead of embedding responsible AI principles from the outset.
Limited imagination, often due to capacity constraints—focusing on email drafting rather than systemic transformation.
For Vilas, the sector is still not thinking big enough about AI’s possibilities.
“Start by asking: what data do we uniquely have that represents the lived experience of the communities we serve?”
“Governance cannot be something we bolt on later. It has to be a design decision from the very beginning.”
6. Cross-Sector Partnerships Are Critical
Vilas calls for deeper connections between technology companies, philanthropies, and nonprofits to share knowledge and co-create socially beneficial AI solutions. He highlights three priorities:
Building stronger bridges between tech companies and the social sector
Creating many-to-many networks for nonprofits to share learnings and collaborate
Developing foresight capacity to anticipate future AI challenges rather than responding reactively
He envisions cross-sector partnerships that bring technical expertise into social impact innovation, expanding the infrastructure for “social-purpose AI.”
7. AI Policy Is Diverging Globally
Vilas warns of a trend where governments abdicate their responsibility to regulate AI in the public interest, leaving policy shaped by speed and competition rather than democratic values.
==
All Tech Is Human is creating a systemic change approach to tackling thorny tech & society issues and aligning our tech future with the public interest. Our wide range of activities allows for broad public engagement and insight, multidisciplinary expertise throughout our global network, and rapid distribution of key ideas and best practices to all stakeholders.