Democratizing AI: Scaling Education Across 12 Countries

The Global AI Skills Crisis

We are living through the most consequential technological shift since the internet, and the world is not ready for it. According to the World Economic Forum, 86% of employers anticipate AI to drive transformation in their organizations in the coming years, while 39% of core workplace skills are expected to change by 2030. The gap between the AI capabilities organizations need and the talent available to deliver them has become one of the defining economic challenges of our era.

The numbers are stark. There are an estimated 1.6 million unfilled AI positions globally [5]. Demand for AI skills exceeds supply at a ratio of roughly 3.2 to 1 across key roles [8]. ManpowerGroup's 2025 Talent Shortage Survey found that 76% of employers struggle to fill jobs due to skill shortages [3], while 94% of business leaders report facing AI-critical skill shortages in their organizations [4]. IDC has projected that these skills shortages could cost the global economy as much as $5.5 trillion by 2026 in lost revenue and operational inefficiency [2].

And here is what I find most troubling: only about one-third of employees received any AI training in the past year [4]. Despite the urgency, despite every headline about AI changing everything, the vast majority of working professionals have not been given a meaningful on-ramp to understand and work with this technology. The supply of education is nowhere close to matching the scale of the demand.

This is the problem I set out to address when I founded Saturdays.AI. Not with another expensive master's degree or an elite Silicon Valley bootcamp, but with a model that could reach people wherever they are -- across languages, across borders, across income levels.

Why Saturdays.AI Exists

The idea for Saturdays.AI came from a simple observation: the people who most needed AI skills were the least likely to have access to them. In 2018, when I was working in technology consulting, I saw firsthand how AI was reshaping industries -- and how unprepared most professionals were for that shift. The few quality AI programs available were prohibitively expensive, geographically concentrated, and designed for people who already had technical backgrounds.

"AI is too important to be left only to engineers and computer scientists. Every professional, every community, every country deserves the chance to understand and shape this technology."

-- The founding principle of Saturdays.AI

I believed that AI education could be delivered differently. That you could build a rigorous, hands-on program and make it free -- or nearly free. That you could train people in their own cities, in their own languages, on Saturdays, without asking them to quit their jobs or take on debt. That you could build a community-driven model that scaled not through venture capital, but through the passion of local ambassadors who believed in the mission.

So in early 2019, we launched the first Saturdays.AI cohort in Madrid. Forty-two people showed up on a Saturday morning to learn about machine learning. They came from backgrounds you would not find in a typical computer science classroom: doctors, lawyers, journalists, educators, small business owners. Within sixteen weeks, every one of them had built an AI project. Several of those projects went on to address real social challenges -- from detecting fake news in Spanish media to predicting air quality in urban neighborhoods.

That first cohort proved the model worked. What happened next proved it could scale.

The AI Skills Gap: By the Numbers

To understand why this work matters, it helps to see the scale of the problem we are up against. The data tells a story of massive, systemic underinvestment in AI literacy.

The economic implications are profound. McKinsey estimates that by 2030, up to 30% of U.S. work hours could be automated by AI-driven technologies [4]. Meanwhile, the PwC Global AI Jobs Barometer finds that jobs requiring AI skills command a 56% wage premium over comparable positions that do not [8]. The people and countries that invest in AI skills now will capture the economic upside. Those that do not risk being left behind.

The Model: How AI Saturdays Works

The Saturdays.AI bootcamp model was designed from the ground up to solve the access problem. Every design decision was made with one question in mind: how do we remove the barriers that keep most people away from AI education?

The program runs for approximately 14 to 16 weeks. Participants learn the fundamentals of machine learning, deep learning, natural language processing, and computer vision through a combination of structured lectures, hands-on labs, and collaborative project work. The curriculum is kept current and practical -- we teach what people actually need to build things, not abstract theory disconnected from application.

Critically, the model is free or very low cost to participants. We partner with universities, coworking spaces, and technology companies who provide venues and resources. Our instructors and mentors are volunteers -- experienced practitioners who give their Saturdays because they believe in the mission. The ambassador model means we can launch a new city without needing to hire staff or build infrastructure. A passionate local leader, a willing venue, and a community of learners -- that is all it takes.

Scaling Across 12 Countries

After Madrid, the growth was organic and, honestly, faster than I expected. Within the first year, we had expanded to Barcelona, Bilbao, and Valencia. Then requests started coming from outside Spain -- from Mexico City, from Bogota, from Lima. People had seen what we were doing and wanted to bring it to their communities.

Today, Saturdays.AI operates across 12 countries. We have run programs in Spain, Mexico, Colombia, Peru, Argentina, Chile, Ecuador, Bolivia, the Dominican Republic, Portugal, France, and the United States. Each city adapts the core model to its local context, but the fundamental principles remain the same: accessible, hands-on, community-driven, and focused on social impact.

Scaling across borders taught us things we could never have anticipated. In some Latin American cities, internet connectivity was unreliable enough that we had to redesign our lab exercises to work with pre-downloaded datasets. In some European cities, data privacy regulations meant we needed to rethink how we handled certain types of project data. In every city, the local ambassador's relationships with universities and companies were the difference between success and failure.

"You cannot parachute a Silicon Valley model into Quito and expect it to work. The curriculum may be universal, but the delivery must be deeply local."

-- Lesson learned from scaling across Latin America

The hardest challenge was quality control at scale. When you have dozens of volunteer-led programs running simultaneously across multiple time zones, maintaining a consistent standard of instruction is genuinely difficult. We addressed this by building a shared curriculum platform, running regular train-the-trainer sessions for ambassadors and mentors, and creating a peer review system where cities could learn from each other's successes and failures.

AI4Good: Where Education Meets Impact

From the very beginning, we made a deliberate decision that every Saturdays.AI project should address a real problem. We call this AI4Good, and it is the heart of what makes this program different from a typical bootcamp. We are not training people to optimize advertising clicks. We are training people to use AI for things that matter.

Over 500 AI4Good projects have been built through Saturdays.AI, and they span an extraordinary range of challenges. Projects have tackled environmental monitoring, early disease detection, educational accessibility, agricultural yield prediction, misinformation detection, and urban mobility optimization, among many others. Some of these began as student projects and evolved into real deployments or even startups.

A few examples that stay with me. In Lima, a team of participants -- none of whom had any prior programming experience -- built a computer vision system to classify types of recyclable waste for informal waste collectors. In Barcelona, a group of healthcare professionals developed a natural language processing tool to help triage patient intake forms in overloaded clinics. In Mexico City, a team built a model to predict water supply disruptions in underserved neighborhoods.

These are not toy projects. They represent what happens when you give real people the tools and confidence to apply AI to the problems they live with every day. The domain expertise comes from the participants themselves. We just provide the technical framework.

The Diversity Challenge

One of the reasons I started Saturdays.AI was to push against the alarming concentration of AI talent and opportunity. The numbers in this space are deeply concerning.

67% of AI talent is concentrated in just 15 major cities worldwide [5]. 89% of AI PhD programs are located in developed countries [5]. Women hold only 28% of AI positions, compared to 51% of the general workforce [6]. These are not just statistics -- they represent a fundamental failure to include the perspectives, needs, and creativity of most of the world's population in the development of the technology that will shape all of our futures.

When AI systems are built by a narrow slice of humanity -- predominantly male, predominantly from wealthy cities in wealthy countries -- those systems reflect a narrow set of experiences and priorities. We see this in biased facial recognition, in language models that work well in English but poorly in Quechua or Catalan, in health AI trained on data from populations that look nothing like the patients it will eventually serve.

Saturdays.AI was designed explicitly to challenge this pattern. By operating in Latin America and Southern Europe, by offering programs in Spanish, Portuguese, and French, by actively recruiting participants from non-technical backgrounds, and by keeping the cost barrier as low as possible, we have been able to reach communities that the mainstream AI education pipeline largely ignores. It is not enough, but it is a start.

What We Have Learned

After years of running Saturdays.AI across multiple continents, a few lessons stand out. These are things I wish someone had told me at the beginning, and they are relevant to anyone trying to scale an education initiative internationally.

Community is the product. The curriculum matters, of course. But what keeps people coming back on Saturday after Saturday is the community -- the sense of belonging, of learning alongside people from different backgrounds who share the same curiosity. Alumni networks have become the most powerful driver of our growth. People recommend us not because of our slides, but because of the experience of being part of something larger than themselves.

Local leadership is non-negotiable. Every successful city we have launched has been driven by a passionate local ambassador. Every city where we struggled had a leadership vacuum. You cannot centralize your way to community. The ambassador model is messy and hard to control, but it is the only model that produces authentic engagement across diverse cultural contexts.

Projects are better teachers than lectures. The single most effective pedagogical decision we made was requiring every participant to build a real project. People retain almost nothing from a sixteen-week lecture series. They retain everything from the experience of struggling to get a model to work on a problem they care about. The project is the learning.

Diversity requires intentional design, not just good intentions. If you build a program and simply hope that diverse participants will show up, they will not. You have to design for inclusion at every level: scheduling, location, cost, language, marketing imagery, the composition of your teaching team, the examples in your curriculum. We have gotten better at this over time, but we are still learning.

Scale and quality are in constant tension. As we expanded from one city to dozens, maintaining quality became our biggest operational challenge. The train-the-trainer model, the shared curriculum, and the peer review systems help, but there is no substitute for strong local leadership. Quality does not scale automatically. It scales through people.

"The AI revolution will not be built by a few thousand researchers in California. It will be built by millions of people around the world who have the skills and confidence to apply this technology to their own contexts. That is the future we are working toward."

-- Miguel Guerrero

Looking ahead, the challenge is only getting bigger. McKinsey estimates that by 2030, up to 30% of U.S. work hours could be automated. The World Economic Forum tells us that 39% of core skills will change. The demand for AI literacy is not a temporary spike -- it is a structural shift in what it means to be a competent professional in almost any field.

Saturdays.AI is one answer to that challenge, but it is far from the only one needed. We need governments to invest in AI literacy at the primary and secondary education level. We need companies to fund serious upskilling programs for their existing employees, not just superficial lunch-and-learn sessions. We need universities to rethink curricula that were designed for a pre-AI world. And we need more community-driven initiatives that meet people where they are, in their languages, in their cities, on their terms.

The work continues. Every Saturday, somewhere in the world, a group of people is sitting down together to learn how machine learning works, to build something with their own hands, to imagine how AI could make their community a little bit better. That is what keeps me going. That is why Saturdays.AI exists.