Table of Contents
The former U.S. CTO reveals how Buffalo's environmental activism shaped her approach to bringing hidden innovators together and why AI needs a "kindness test" instead of just intelligence metrics.
Key Takeaways
- Innovation works best through "community organizing" - finding and connecting existing doers rather than starting from scratch
- The "glue layer" is the most underfunded area in tech - the ecosystem support for surfacing and connecting innovators worldwide
- President Obama's response to AI warnings was "let's get started" - leading to pioneering public engagement on AI policy in 2015-2016
- Universities should teach interdisciplinary "tools" - from coding to drama to storytelling - because "the universe doesn't ring a bell between classes"
- Hidden figures like Ada Lovelace, Grace Hopper, and George Washington Carver show how women and minorities have always driven technological breakthroughs
- Half of MIT students are now in computer science, but we need engineers working on environmental and social problems, not just data science
- Government is "only whoever shows up" - making it critical for technologists to rotate into public service
- The key question for AI isn't intelligence but kindness - can we build systems that help with justice, peace, and reducing bullying?
- Buffalo's innovative teachers created mandatory science fairs where every student had to invent something related to their passions, not "baking soda volcanoes"
From Buffalo's Burning River to Global Innovation Networks
Megan Smith's approach to innovation didn't start in Silicon Valley boardrooms or MIT labs. It began in 1970s Buffalo, where her city's river was literally on fire from industrial pollution, the schools were being integrated through pioneering magnet programs, and her mother started a bicycle club on the first Earth Day.
This wasn't just environmental activism - it was a masterclass in community organizing that would later revolutionize how America thinks about technology and innovation. Smith's teachers at her cash-strapped public school had turned educational limitations into creative freedom, designing experiences that mixed academic learning with real-world problem-solving.
The most transformative was their mandatory science fair policy. Unlike typical school projects, students couldn't make "baking soda cookie volcanoes." Instead, teachers asked each child: "What do you care about?" Then they had to discover or invent something related to that passion. This approach, which Smith calls "practice makes permanent," launched countless STEM careers because students learned they could actually do science and see how to apply it.
One year, teachers organized something called "1984 day" where they started implementing Orwellian policies without telling students, waiting to see when kids would revolt. It took until after lunch before students organized a town hall meeting. Another time, they created "ideal community" week where teachers remixed their subjects around their own passions, leading to architectural tours of Buffalo's Frank Lloyd Wright buildings and ecosystem studies that included biking to sewage treatment plants.
This wasn't just creative pedagogy - it was training in what Smith now calls "community organizing for innovation." The key insight: don't just think about companies as innovators. Think about water treatment facilities, transportation systems, and all the built infrastructure around us. Who are the innovators working in those spaces, and how do we connect them?
That Buffalo foundation shaped everything that followed, from her work at General Magic creating technologies that became the iPhone, to her role as U.S. CTO bringing this inclusive innovation philosophy to government, to her current work at Shift Seven connecting global problem-solvers.
The Obama AI Strategy: Open Everything Up
When Megan Smith sent President Obama a memo in 2016 explaining that narrow AI was advancing rapidly but general AI was still decades away, she expected a typical political response - maybe form a committee or commission a study. Instead, Obama wrote back: "That's not all that far off. Let's get started."
But Obama's "let's get started" wasn't about racing to build better AI or creating secret government programs. It was about opening up the conversation to the American people in unprecedented ways. Working with Ed Felten and Terah Lyons, Smith's team designed a series of public town halls at universities across the country, fully open to anyone who wanted to participate.
At University of Washington, they focused on law and policy. At Carnegie Mellon, it was AI safety and control. At NYU, they partnered with the Council of Economic Advisers to explore impacts on jobs and wages. Each event brought together technical experts, social scientists, community organizers, and ordinary citizens who were curious or concerned about AI's implications.
The strategy revealed something crucial that's still missing from today's AI discussions. When people talk positively about AI, Smith noticed, they always mention healthcare improvements or productivity gains. But there's a "huge vacuum" when it comes to applying AI to social safety nets, environmental justice, or reducing inequality. Why isn't there AI focused on getting food to everyone in cities, reducing bullying, or creating more peace and justice?
This wasn't just about managing AI risks - it was about democratizing the conversation about what AI should actually do. Smith wanted to "crowdsource the range of perspectives, the range of fears, the range of enthusiasm" rather than letting the technology's direction be determined solely by whoever had the most resources to build it.
The work continued across administrations, laying groundwork for Biden's AI Bill of Rights and executive orders. But Smith's broader point remains: we're building incredibly powerful technology while having remarkably narrow conversations about its purposes. The technical capabilities keep advancing, but the social imagination about beneficial applications hasn't kept pace.
The Missing "Glue Layer" That Connects Hidden Innovators
If Smith had $100 million to invest in solving the world's problems, she wouldn't fund new startups or research labs. She'd invest in what she calls the "glue layer" - the ecosystem support for finding and connecting innovators who are already working but lack resources and visibility.
This philosophy emerged from her work on Google Earth and Maps, where her team needed metadata about street names worldwide. Instead of sending Google employees to collect this information, engineer Lalitesh Katragadda created MapMaker - essentially a Wikipedia for geography where local people could draw their own communities onto satellite imagery.
The results were extraordinary. In Lahore, Pakistan, residents mapped their entire city in six months. Across the developing world, communities that lacked resources for official surveying suddenly appeared on digital maps, making it possible for ambulances to find addresses and businesses to receive deliveries.
This revealed a fundamental truth: there's extraordinary talent and innovation happening everywhere, but most of it remains invisible to traditional funding and support systems. Smith started documenting examples: someone flying drones to plant a billion trees annually, people in Uganda teaching law in prisons, floating fabrication labs in the Amazon that let indigenous communities participate in vaccine development without destroying their forests.
The UN Solutions Summit that Smith helped organize found innovators like Bernice, who builds bamboo bicycles in Nigeria, and discovered that Boise, Idaho has fifteen tech meetups with one drawing 800 people - but most Boise residents don't know this ecosystem exists.
The pattern is consistent: solutions to major problems already exist, created by people who understand those problems intimately. What's missing isn't more innovation - it's better systems for finding these innovators, connecting them with each other, and providing the support they need to scale their impact.
This "glue layer" approach represents a fundamental shift from the typical Silicon Valley model of identifying problems and building solutions from scratch. Instead, it's about recognizing that the people closest to problems often have the best insights about solutions, and focusing investment on amplifying their work rather than replacing it.
Why Universities Need to Stop Ringing Bells Between Classes
When Smith advises students about what courses to take, her guidance is deceptively simple: follow your passion, learn as many tools as possible, and remember that "the universe doesn't ring a bell between classes."
That last point captures something crucial about how education needs to evolve. Universities create artificial boundaries between disciplines, but real-world problems don't respect those boundaries. Climate change requires chemistry, policy, engineering, economics, psychology, and storytelling. Criminal justice reform needs data science, sociology, law, community organizing, and user experience design.
Smith's favorite example is Joy Buolamwini's "evocative audit" called "AI Ain't I a Woman" - a spoken word poem with visuals that demonstrates how facial recognition systems fail for women and people of color. This work required technical expertise to analyze algorithmic bias, artistic skills to create compelling performance, and deep understanding of social justice history to connect contemporary AI problems to Sojourner Truth's famous speech.
When Buolamwini performed this piece at Lincoln Center, you could hear a pin drop. A traditional academic paper might reach hundreds of researchers, but the artistic presentation reached thousands and created emotional understanding that pure data couldn't achieve.
This cross-pollination of skills becomes more important as problems get more complex. Smith points to Ellen Swallow Richards, MIT's first female graduate, who in the 1800s started testing water quality in the Charles River and essentially invented environmental engineering by combining chemistry with public health concerns.
For today's students, Smith recommends breadth over premature specialization. Take a drama class to improve public speaking. Learn data science and storytelling. Study how Jane Addams created "settlement houses" in Chicago neighborhoods that combined academic knowledge with community organizing. Understand how Grace Hopper invented programming languages because she wanted to "broaden participation in computing" by making code more human-readable.
The goal isn't to become superficial in everything, but to develop enough fluency across disciplines that you can collaborate effectively with experts in other fields. Most important innovations happen at the intersections between domains, and universities need to prepare students for that reality.
Finding Hidden Figures Who Shaped Our World
One of Smith's most important contributions has been systematically uncovering the "hidden figures" - especially women and people of color - whose innovations shaped our technological world but were erased from mainstream narratives.
Take Ada Lovelace, who wrote the first computer algorithm in the 1840s and predicted that machines could eventually understand "the math of the cerebrum" - essentially calling for artificial intelligence. Lovelace was working during Darwin's era, combining mathematical brilliance with artistic sensibility (she loved music and flight). When Google created a doodle for her birthday, "Ada Lovelace" trending globally on Twitter for 20 hours, introducing millions to her story.
Or consider Susan Kare, who designed every graphical interface element we use today - icons, fonts, the visual language of personal computing. Yet in the Hollywood movie about Steve Jobs, her contribution is reduced to one line: "Susan did the bag." As Smith notes, maybe Kare designed a bag, but she also "designed the graphic user interface for all of humanity."
George Washington Carver represents another type of hidden figure - widely known as "the peanut guy" but actually the scientist who saved American agriculture from total collapse. Carver invented crop rotation techniques that restored nitrogen to soil while maintaining profitability. Einstein called him "the great scientist George Washington Carver" and said everyone should know his work. There's still a photo of Carver above the Secretary of Agriculture's desk.
These erasures aren't accidental. They reflect systematic biases about whose contributions get remembered and whose get minimized. When Smith's sister-in-law wrote about "erasable humans," she captured how certain people's achievements simply evaporate from historical memory while others become legendary.
Understanding these hidden figures isn't just about historical accuracy - it's about recognizing patterns of innovation that can inform current work. Lovelace's interdisciplinary approach, combining math with music and art, reflects the kind of boundary-crossing that produces breakthrough innovations. Carver's focus on sustainable systems that work for both environmental and economic needs offers models for addressing climate change.
Most importantly, these stories demonstrate that transformative innovation has always been more diverse and inclusive than mainstream narratives suggest. The challenge isn't bringing diversity to technology for the first time - it's recovering and building on diverse traditions of innovation that have always existed.
The Kindness Test for Artificial Intelligence
When asked what criteria a "Megan Smith test" for AI would include - analogous to the Turing test for intelligence or the Lovelace test for creativity - Smith's answer was immediate: kindness.
"Kindness is so much more important than knowledge, even though knowledge is important," she explained. This isn't soft-hearted sentimentality - it's a fundamental reframe of what we want AI systems to accomplish.
Current AI development focuses heavily on capability: can systems process language, recognize images, play games, write code? But Smith argues we're asking the wrong questions. Instead of "Can this system pass for human intelligence?" we should ask "Does this system make the world more kind?"
This connects to her broader concern about AI development priorities. When technologists discuss positive AI applications, they typically mention healthcare diagnostics, productivity improvements, or business optimization. But there's a massive gap when it comes to applying AI to social safety nets, environmental justice, reducing bullying, or ensuring everyone in a city has access to food.
The imbalance isn't accidental. There are enormous financial incentives for AI that increases corporate profits or replaces expensive human workers. But there's no comparable funding mechanism for AI that reduces inequality or strengthens communities.
Smith's "kindness test" would evaluate whether AI systems actively contribute to human flourishing and social justice. Can this system help detect and reduce bias rather than amplifying it? Does it strengthen community connections or isolate people further? Does it make resources more accessible to people who need them most?
This approach requires what Smith calls "crowdsourcing" the criteria - involving communities who would be affected by AI systems in defining what beneficial outcomes actually look like. Rather than having tech companies or government agencies decide unilaterally what "good AI" means, the standards should emerge from inclusive conversations with diverse stakeholders.
The urgency comes from recognizing that AI systems trained on biased historical data will reproduce and amplify those biases at unprecedented scale. We're not just building neutral tools - we're encoding particular values and assumptions into systems that will shape millions of decisions. The question is whether we'll be intentional about what values we're encoding.
Why Government Needs More Technologists (And Why Technologists Need Government)
Smith's years as U.S. CTO convinced her that one of the biggest barriers to solving complex problems is the artificial separation between technological capability and public purpose. Government has the mandate and resources to address society's biggest challenges, but often lacks the technical expertise to do it effectively. Meanwhile, tech talent concentrates in companies focused on advertising optimization and engagement metrics rather than education, healthcare, or climate change.
Her solution is systematic rotation of technologists through government service, similar to how lawyers clerk for judges or business leaders take roles in federal agencies. Programs like the U.S. Digital Service, 18F, and Presidential Innovation Fellows now create pathways for technologists to spend 1-2 years working on public problems before returning to private sector careers.
The impact can be transformative. Smith describes working with amazing public servants who deeply understand policy and implementation challenges but need technical partners who can rapidly prototype solutions and navigate modern development practices. Government problems are often harder than commercial ones because they involve multiple stakeholders, complex regulations, and populations that can't be ignored if they're not profitable.
But the learning goes both directions. Technologists working in government gain appreciation for the complexity of democratic processes and the importance of building systems that work for everyone, not just early adopters. They see how technical decisions become policy decisions with real consequences for vulnerable populations.
Smith's broader point is that "government is only whoever shows up." If technologists don't engage with public service, then technological governance gets shaped by people who may not understand the systems they're regulating. This creates a dangerous feedback loop where technology policy lags behind technological reality, leading to either ineffective oversight or reactive regulations that stifle beneficial innovation.
The urgency has increased with AI development, where decisions made in the next few years will shape how these systems integrate into everything from healthcare to criminal justice. Having technologists who understand both the capabilities and limitations of AI systems involved in policy development isn't optional - it's essential for ensuring those policies actually work.
The Global Network of Local Solutions
Smith's most ambitious vision involves scaling the "glue layer" approach globally - creating networks that connect local innovators working on similar problems in different contexts, allowing solutions to spread and adapt across communities.
The model emerged from her work tracing nuclear fallout from the 1945 Trinity test, where she connected Princeton researchers with community organizers in New Mexico to analyze weather data and understand how radioactive particles spread across 46 states. This wasn't about creating new technology - it was about applying existing capabilities to an environmental justice problem that had been ignored for decades.
The same approach works for contemporary challenges. When Smith asked the global community "who's already solving the Sustainable Development Goals," the responses revealed an extraordinary ecosystem of local innovations: floating fabrication labs that let Amazon communities participate in vaccine development, women's textile cooperatives in Nigeria, law education programs in Ugandan prisons, drone-based reforestation projects.
Each solution is deeply rooted in local context and community knowledge. But many face similar technical or organizational challenges that could be addressed through cross-pollination. The bamboo bicycle workshop in Nigeria might benefit from connecting with sustainable transportation advocates in Boise. The reforestation drone team could share techniques with urban air quality monitors in West Virginia.
This global network approach avoids the colonial implications of "scaling" successful Western innovations to developing countries. Instead, it recognizes that communities closest to problems often develop the most effective solutions, and focuses on creating infrastructure for sharing knowledge and resources across these distributed innovation networks.
The vision extends beyond technology to governance models. Smith points to the contrast between Burning Man and refugee camps - similar scale and budget, but radically different approaches to community organization. One creates conditions for self-actualization and creative expression; the other imposes top-down service delivery that limits autonomy and agency.
Understanding these differences offers insights for designing better systems everywhere - from how cities organize emergency response to how schools structure learning environments. The goal isn't copying specific solutions but understanding underlying principles about human agency, community self-organization, and institutional design that can adapt across contexts.
