Somewhere in rural England, America, or really anywhere rural, there is a classroom where worksheets are still being copied by hand. At a nearby private school with lots of money, a coworker uses AI to make personalized lesson plans, write letters to parents, and make tests for students in a lot less time. They are both trying to do the same thing. Most of the people in charge aren’t paying nearly enough attention, but the gap between them is getting bigger.
It’s hard to take your eyes off the numbers. The Sutton Trust in the UK did research that showed teachers in private schools are more than twice as likely as teachers in public schools to have had formal training in AI (45% vs. 21%). It was found that state schools serving students from wealthier neighborhoods had much higher rates of both formal and informal AI training than schools serving students from poorer neighborhoods. There is a big gap. The gap keeps getting bigger.
The timing of this makes it even more uncomfortable. There is still a lot that schools need to learn about how AI can be used in the classroom. Rules are being written down. Getting into habits. During this window, the person or organization that sets up the right training and infrastructure will have a structural advantage for years to come. It will take that person the same amount of time to catch up, which rarely goes as planned when it comes to technology.
It’s not just that private schools train their teachers better. Also, they’re using AI in new ways. In schools with lots of resources, teachers use these tools for a lot of different tasks, from grading and assessing to continuing their professional development. When schools don’t have enough money, AI is used in a narrower and less consistent way. There’s a real chance that schools with more resources will use AI to really help their teachers, while schools with less resources will only use it to cover up gaps they can’t otherwise fill. That’s not fair. That’s not the same problem as the label says.
A big part of the problem is infrastructure, which is something that most policy discussions conveniently skip over. Things like fast internet, new computers, and tech support staff cost money, and schools in rural areas and with low incomes tend to have less of it. Old hardware doesn’t work well with AI tools. Generative platforms that use large datasets often reflect the cultural assumptions and priorities of the places that funded them. This means that students from all walks of life have to look for content that speaks to them personally. A student in a poor urban school or a school in the middle of nowhere in the country isn’t just missing a login. They are missing out on a whole educational system that their wealthier peers don’t think twice about.
It’s still not clear if governments are taking this seriously enough to make it an emergency. There are good ideas in the Sutton Trust’s suggestions: give all eligible low-income students devices, make every school appoint a senior staff member to oversee AI, actively look for gaps between schools, and put money toward closing them. The ideas here aren’t very radical. They are basic rules for building things that are wrapped up in modern language. As always, it’s a different question whether there is political will to follow through.
Seeing how everything is going makes me think that the AI moment in education could go one of two ways. It could become the great equalizer, giving teachers who are already busy more time, giving students who are interested more access, and giving schools that don’t have enough resources the tools they need to do more with less. It could also do what many other waves of educational technology have done: it could come with a lot of promise, but the students who are already ahead would take it in the fastest, leaving the weakest students to watch from afar. The technology doesn’t make the choice. The money does. The policy does. Will in politics does. All of those things are moving too slowly right now.
