Editor’s note: This essay is an entry in Fordham’s 2026 Wonkathon, which asked contributors to answer this question: “What rules should govern the uses of digital technology in schools—and who should set them?” Learn more.
In his 1845 Narrative, Frederick Douglass recalls the day his enslaver caught his own wife teaching Douglass to read and told her it would “forever unfit him to be a slave.” From then on, Douglass “understood the pathway from slavery to freedom.”
That freedom starts with comprehension: building a mental model of what an author describes and connecting it to what you already know. These skills are slipping. U.S. reading scores fell 14 points between 2022 and 2025 on PISA, and the share of “hasty readers” who read quickly but inaccurately has nearly doubled since 2018.
Meanwhile, AI is changing what literacy means. AI predicts likely words and patterns and has no stake in their truth. Making sense of the output must now be part of literacy. It’s going to take more discernment, judgment, and critical thinking than ever. AI offers the illusion of credibility; only when you dig beneath the surface do you realize it requires significant background knowledge, critique, and human reasoning to produce anything worthwhile. It’s called AI slop for a reason.
Tech opened information to students who lacked access to it. But we rolled it out fast, with little planning. Now parents, AI leaders, and students are calling for limits. The capacity to think freely and critically grows over time and needs care and protection while it takes root. We’ve written rules around what we want to prevent: distraction, scrolling, cheating, AI companions. We’d do better to start with what we want to protect. If literacy is freedom, kids need three things to become free readers and critical thinkers: attention, relationships, and trust.
Protection #1: Attention
Deep reading and rigorous thinking take sustained focus. Protecting it means guarding students’ attention while it develops.
- Print and people first. A poor reader at the end of first grade has an 88 percent chance of remaining one in fourth grade. Children learn to read with a book in hand and an adult beside them, and screens can get in the way: parents reading e-books with toddlers talk and collaborate less than with print. Older students need practice with long, demanding texts, and paper helps: a meta-analysis found readers understood informational texts better on paper, especially under time constraints.
- States should set print-first expectations, especially for early grades and long-form reading.
- Districts should define exceptions, such as assistive technology and English learner support.
- Create a perimeter but expand the fence over time. The skills behind sustained focus develop rapidly from ages 10–15 and mature around 18–20; many apps rely on features that “prey on developing brains.” But students who never learn to apply critical thinking and literacy to these tools will leave school unprepared.
- In elementary school[1], most students shouldn’t spend time on screens, and shared devices should serve instruction. This is also the time to start talking with kids about what AI is.
- In middle school, districts should block games, social media, and streaming from school devices.
- In high school, districts should loosen settings in stages and involve students in setting the rules. They should train teachers to help students read AI output just like any other source, checking claims, and deciding when to do the thinking themselves.
Protection #2: Relationships
The science of learning shows that relationships are foundational for learning, and dialogue helps critical thinking grow. Discussing texts in class improves comprehension; questioning them together builds critical thinking. Middle schoolers who practiced arguing with peers developed stronger reasoning. All of this depends on friction that chatbots are designed to remove.
- Find the friction. Low-tech, discussion-centered work should be the standard. When schools do use tech, districts should require tools that default to prompting students to explain, work with others, question, and revise their thinking.
- No fake friends. According to Sherry Turkle, “Social media came for our attention and chatbots come for our capacity for attachment.” Common Sense Media reports that 72 percent of teens have used AI companions; a third use them for social interaction and relationships.
- School systems can hold vendors to the new AFT, UFT, and Microsoft standard, which bans companion features.
- States can lead. California now requires safeguards for minors.
- And make room for real ones. Lunch, hallways, and recess are where kids practice being part of a community.
- Personal devices should stay out of schools[2]. When Norwegian schools restricted smartphones, bullying fell.
Protection #3: Trust
Over 20 years ago, Anthony Bryk and Barbara Schneider showed how schools depend on trust. That trust is under strain. Schools are “bracing for a growing backlash,” and the people closest to kids are worried AI harms critical thinking. To many families, this feels like an experiment being run on their kids.
Adults need to exercise the same discernment, judgment, and critical thinking we ask of kids. That means bringing communities into decisions and holding vendors accountable for results.
- Show families the whole picture. Families should know which tools their children use, what data they collect and how they affect learning.
- We cannot trust what we cannot see. Districts should use the AFT, UFT, and Microsoft standard to require vendors to publish plain-language guides for families.
- Districts should bring communities into decision-making. Listening sessions and technology councils are a start.
- Make honesty easier than hiding. AI has made cheating easier. It’s tempting to respond with tighter control. But a new study found teens whose parents trusted them to learn to manage technology reported more self-regulation. AI detectors are unreliable, too. For example, they overidentify work by non-native English writers as AI-generated.
- Districts should write clear rules with students that spell out when AI is allowed and when it isn’t, and they should stop treating detector results as proof of cheating.
- Teachers should build in work that shows thinking as it happens, like in-class drafts and short conversations about a student’s work.
- A student who admits to misusing AI should be given a chance to redo the work, while one who hides it faces consequences.
- Hold vendors accountable. Most education technology lacks evidence. Some districts are using outcomes-based contracts with promising early results. But districts should not be left alone to put rules in place. Some states have started to lead. Utah funds math software only from an approved list with evidence; Texas has adopted a similar approach. But if every state builds its rules, we risk a patchwork with no way to compare results. And we need much more scrutiny on the methodology for evaluating these tools.
- Funders should resource new methods, evaluations, and a public record of which tools meet common standards.
Love, actually
Getting these protections right will require reading, thinking, and talking with other people to decide what belongs in schools. This looks like adults who guard kids’ focus and trust them with more freedom as they grow, and people who push back on their ideas. What holds them together is a form of love: a steady effort toward collective growth. That’s how free people govern themselves, and it gives students what Douglass got from reading: the ability to question the world others have built and choose their own path through it.
Literacy is freedom, and our school technology rules should protect it.
