By Dr. Janette Camacho | October 28, 2025

Two years into the generative AI era, we have enough data to move beyond anecdote. And the data tells a story that should alarm every educator, policymaker, and parent who cares about educational equity: AI literacy is stratifying along exactly the same demographic lines that have defined American educational inequality for generations. Income. Race. Geography. School funding. The students who need AI competency most to compete in a transformed economy are the students least likely to receive it.

This article presents the evidence, traces the causal mechanisms, and argues that without deliberate, funded intervention, AI will become the latest technology to widen the gaps it was supposed to close.

The Numbers

A September 2025 report from the Pew Research Center found that 72% of students in households earning above $100,000 annually reported using AI tools for schoolwork at least weekly. Among students in households earning below $40,000, that figure was 29%. The gap was not primarily about access to devices or internet - though those barriers persist - but about something subtler: whether students had been taught to use AI tools productively and whether their schools treated AI as a skill to develop or a risk to manage.

The RAND Corporation's spring 2025 teacher survey provided the supply-side data. In high-poverty schools (defined as those where 75% or more of students qualify for free or reduced-price lunch), only 18% of teachers reported integrating AI tools into instruction. In low-poverty schools, the figure was 54%. The gap was even wider for schools in rural areas, where 12% of teachers reported AI integration compared to 47% in suburban schools.

The National Assessment of Educational Progress (NAEP) does not yet include AI literacy as a measured domain, but the 2024 Technology and Engineering Literacy assessment provided suggestive data. Students in the highest socioeconomic quartile scored 34 points higher than those in the lowest quartile on items assessing computational thinking and digital problem-solving - competencies that underpin AI literacy. That 34-point gap represents approximately two grade levels of difference.

These are not surprising numbers. They mirror patterns we have seen with every educational technology from personal computers to broadband internet. But they should be infuriating, because this time, we saw it coming. We had time to act. And we largely did not.

The Three Gaps

The AI literacy divide is not a single gap. It is three interconnected gaps that compound one another.

The Access Gap

Despite improvements in device availability since the pandemic-era one-to-one initiatives, meaningful AI access remains unequal. Access means more than having a Chromebook. It means having a device powerful enough to run AI-intensive applications, internet bandwidth sufficient for cloud-based tools, and school network policies that do not block AI platforms entirely.

A 2025 CoSN (Consortium for School Networking) infrastructure survey found that 31% of school districts - disproportionately rural and high-poverty - still have internet bandwidth below the FCC's recommended minimum for digital learning. In these environments, web-based AI tools load slowly, time out, or are simply inaccessible. Students who can only use AI at the school library during designated hours do not develop the same fluency as students who practice at home on personal devices.

Additionally, many schools responded to the initial AI disruption of 2023 by blocking ChatGPT, Gemini, and similar tools on school networks. While some have since reversed those decisions, a significant number have not. When schools block AI, the students who lose access are those who depend on school for their technology use. Students with devices and internet at home simply use AI at the kitchen table. The ban creates a de facto two-tier system.

The Instruction Gap

Even when access is equal, instruction is not. The teachers most likely to integrate AI into their practice are those who have received training, have supportive administrators, and have the professional confidence to experiment with new tools. These conditions cluster in well-resourced schools.

I have trained over 500 educators through iTeachAI. The demographics of who seeks out AI training are revealing. Approximately 70% of participants in my workshops come from suburban districts. Roughly 20% come from urban districts. Fewer than 10% come from rural districts. This is not because rural and high-poverty teachers are uninterested. In my experience, they are deeply interested. They are also more likely to face barriers: less flexible schedules, fewer substitutes to cover classes during PD days, less administrative support for innovation, and more immediate competing priorities (basic literacy, staffing shortages, student trauma).

The instruction gap is also a curriculum gap. Schools that integrate AI into instruction tend to do so across content areas - AI in writing, AI in science, AI in math, AI in art. Schools that do not integrate AI leave students to encounter it only in their personal lives, without pedagogical context or critical evaluation skills. The difference is between students who learn to use AI thoughtfully and students who learn to use AI haphazardly.

The Critical Literacy Gap

This is the most concerning dimension and the hardest to measure. Critical AI literacy - the ability to evaluate AI output for accuracy, bias, and manipulation; to understand how AI systems work at a conceptual level; to reason about the societal implications of AI deployment - is a competency that requires intentional instruction.

Preliminary data from a fall 2025 pilot of an AI literacy assessment developed by researchers at UCLA's Center for Digital Literacy found significant disparities in critical evaluation skills. Students who had received structured AI instruction (at least 10 hours across the school year) scored 2.3 standard deviations higher on items assessing the ability to identify AI-generated misinformation than students who had received no instruction. The effect size is enormous and suggests that critical AI literacy is highly teachable - but only if it is taught.

Students without critical AI literacy training are not abstaining from AI. They are using it uncritically. They accept AI-generated text as authoritative, share AI-generated images without questioning their authenticity, and make decisions based on AI recommendations without evaluating the underlying logic. In an information environment increasingly saturated with AI-generated content, this lack of critical literacy is not merely an educational gap. It is a civic vulnerability.

Who Is Being Left Behind

The data points to specific populations at highest risk.

Students in rural communities. Lower bandwidth, fewer trained teachers, smaller districts without technology leadership, and geographic isolation from professional development opportunities create compounding barriers.

Black and Latino students. These students are disproportionately enrolled in high-poverty schools with the lowest rates of AI integration. The Digital Equity Act of 2021 allocated funding to address digital divides, but implementation has been slow and AI-specific provisions are minimal.

Students with disabilities. As I wrote in my May 2025 article, special education has been slowest to adopt AI tools despite having the most to gain. Privacy concerns, funding silos, and risk aversion in a litigated space keep AI out of the hands of students who need it most.

English Language Learners. While AI translation and language support tools have significant potential for ELL students, adoption requires teachers who understand both the tools and the linguistic needs of their students. ELL-serving teachers are among the least likely to have received AI training.

What Must Change

I am not interested in documenting a problem for the sake of documentation. The question is what we do about it. I see three essential interventions.

1. Targeted AI PD Funding for High-Need Schools

Federal and state professional development funding should prioritize AI training for teachers in Title I schools, rural districts, and schools serving high percentages of students with disabilities and English learners. The current allocation model - where PD dollars flow disproportionately to schools that already have capacity - reinforces the gap rather than closing it.

2. AI Literacy Standards with Equity Mandates

ISTE's updated standards are a start, but voluntary adoption is insufficient. States should adopt mandatory AI literacy standards with explicit equity provisions - requiring that assessment data be disaggregated by demographic group and that districts demonstrate equitable access to AI instruction.

3. Community-Based AI Literacy Programs

Schools cannot do this alone. Libraries, community centers, afterschool programs, and faith-based organizations can all provide AI literacy programming that reaches students outside of school hours. The public library system, in particular, has the infrastructure and the mission alignment to serve as a hub for community AI education.

The Stakes

Let me be direct about what is at risk. The U.S. Bureau of Labor Statistics projects that AI-related skills will be relevant to approximately 60% of new jobs created between 2025 and 2035. Students who graduate without AI literacy will face a labor market that was not built for them. They will be consumers of AI systems they do not understand, workers displaced by tools they were never taught to use, and citizens unable to evaluate the AI-driven decisions - in hiring, lending, policing, healthcare - that shape their lives.

We have been here before. The digital divide of the 1990s, the broadband gap of the 2000s, the device gap of the 2010s - each produced inequities that took decades to partially remediate, at enormous social cost. We swore we would not let it happen again. It is happening again.

The AI literacy gap is not inevitable. It is a policy choice, a funding choice, and an instructional choice. Every district that blocks AI without providing alternatives, every state that declines to fund AI professional development, every school that treats AI literacy as enrichment rather than core curriculum is making a choice about which students' futures matter.

I choose to believe we can make different choices. But belief without action is insufficient. The data is clear. The time for intervention is now.

Dr. Janette Camacho is a Google for Education Certified Trainer & Coach, Google Certified Educator Level 1 & 2, Adobe Creative Educator, Apple Teacher, FETC 2024 and 2025 Featured Presenter with 28+ years of K-12 classroom experience. She is the founder of iTeachAI.