Why AI needs to be for everyone, not just tech giants

The ability to create and mold AI is currently locked inside a small number of regions and elite institutions. Most countries and schools act as passive users rather than architects of these systems. This creates a dependency, where schools or entire nations must accept the logic, biases, and goals baked into AI tools built in a different country or by a company with different priorities. Some groups are fighting this trend. For example, at North Carolina Central University, Professor Siobahn Day Grady launched an institute to bring AI education to students at a historically Black university. The goal is to move beyond just using AI, ensuring educators and students have the skills to understand, critique, and eventually build their own systems.

AI requires immense amounts of computing power, usually via giant data centers. Because building and running these centers is incredibly expensive, the infrastructure for the most powerful AI systems is clustered in a handful of countries, primarily the United States. This means that if you are a person or even a government agency in a different part of the world, you cannot build your own AI from scratch. Instead, you access a pre-built model through a company's cloud service. These models perform well for the people who programmed them, but they often struggle to account for local languages, societal norms, or unique regional challenges. When you use these systems, you are using someone else's definition of how an intelligent agent should behave.

The danger is that AI stops being a tool we shape and becomes a gatekeeper we serve. If we don’t expand who has the training and the physical resources to work with AI, we risk a future where a few boardrooms decide which problems are worth solving and whose data matters most. The goal shouldn’t just be getting AI into more schools; it should be making sure that different communities—from local fishermen in Indonesia to students at a wide range of universities—have the power to influence how these systems function. AI is far from finished. The version that exists today is just an early draft. Whether that draft eventually serves everyone depends on making the table where these decisions are made much, much larger.