AI literacy should come before AI adoption. You can’t ask teachers to integrate AI tools into their classrooms if they don’t understand the technology’s foundations, limitations, and ethical stakes. That argument keeps getting reinforced by research, and this study from Hollands and Breazeal (2024) is one of the strongest practical examples I’ve seen of what happens when you actually give educators the resources to teach AI concepts from the ground up.
The paper describes MIT’s “Day of AI” initiative, a set of 12 modular curricula designed to introduce K-12 students and their teachers to core AI concepts. Built in collaboration with an implementation partner called i2Learning and released under a Creative Commons license, the materials were freely available to anyone willing to teach them. The data comes from 2022-23, when 190 teachers used these curricula with approximately 12,000 students across 136 countries. And what it shows is encouraging, if incomplete.
What the AI Literacy Curriculum Covers
What I find compelling about this design is the low barrier to entry. The materials required only basic digital devices and internet access. Teachers with zero AI teaching experience could pick them up. The average preparation time was about four hours, though Hollands and Breazeal note the range was wide: some teachers spent no time preparing, and others invested up to 20 hours adapting content, translating it, or folding in supplementary resources. The most common adaptation was condensing or omitting activities to fit available class time.
I wrote recently about how Chee, Ahn and Lee (2025) built an AI literacy competency framework with progression levels across learning stages, and the Day of AI curricula serve as a concrete example of what that kind of progression looks like when it’s put to work in real classrooms. The K-2 modules ask students to identify AI in everyday objects. The high school modules ask them to draft school AI policies and train their own machine learning models. That’s a genuine developmental arc, built into freely available materials that any teacher in the world can download.
The Numbers Are Promising, With Caveats
Hollands and Breazeal report that teachers’ self-reported AI knowledge rose from 3.8 to 6.0 on a 0-to-10 scale after teaching the curricula. Student knowledge (as reported by teachers) increased from 2.4 to 4.3. Optimism about AI’s potential benefits to society went from 5.6 to 7.1 for teachers and from 5.0 to 6.7 for students. Over 70 percent of implementers said the curricula helped students understand how AI is currently used, how it works mechanically, and its potential to benefit society.
I want to name a limitation here. These are self-reported numbers with no control group. The gains could partially reflect teachers’ growing comfort with the material through repeated exposure, or a novelty effect among students encountering AI concepts for the first time. The authors describe this as formative, pilot-stage research, and that’s a fair characterization.
Still, the direction is consistent enough to take seriously. And one finding struck me as particularly telling: 77 percent of respondents felt they could contribute to the future of AI literacy, but fewer felt the same about shaping the future of AI itself. That gap says something about where teachers feel their real influence lands. They gained confidence in their capacity to educate others about AI, even when the technology itself still felt beyond their direct control. As one teacher put it, “Students are understanding how AI is being used and created and they are understanding that they can become programmers and creators of this technology and not just passive users of it” (p. 41).
When Students Become Policy Thinkers
The most interesting findings aren’t the scale scores. They’re the qualitative ones. Hollands and Breazeal describe a curriculum activity where students collaboratively drafted a school policy on ChatGPT use. The assignment required concrete, understandable guidelines for all stakeholders. One ninth-grader wrote that AI tools should be acceptable for outlines, feedback, and inspiration, but that students shouldn’t copy whole essays or pass off AI-generated ideas as their own.
That’s a nuanced position from a fifteen-year-old. And it tracks with what I’ve seen across the research I’ve covered on this blog. When you give students the chance to think critically about AI, they often arrive at reasonable, balanced positions on their own. The UNESCO AI Competency Framework for Students (2024) describes exactly this kind of ethical reasoning as a core competency, and here is a concrete example of students doing it without being told what to think.
Another student favorite was “AI or Not,” an activity where students examined everyday items like toasters, face filters, and self-driving cars and applied five questions to determine if each qualifies as artificially intelligent. Simple as it sounds, it forces the kind of conceptual thinking about what AI actually is that most adults haven’t done either. Su, Ng and Chu (2023) made a similar argument about starting AI literacy in early childhood, and the Day of AI materials prove that approach can scale across every grade level, with the right scaffolding.
Reading This Data in 2026
I need to contextualize the timing of this study. The data comes from 2022-23, a period when ChatGPT had just launched and the field was still oscillating between panic and hype. Hollands and Breazeal were describing a moment when most teachers had never heard the term “large language model” and AI literacy materials barely existed in K-12 settings.
We are in a different place now. AI is embedded in mainstream tools. Schools have had years of direct experience with these technologies. The research base has expanded enormously. The Day of AI curricula were a first-generation response to an urgent need, and they clearly did real work in that moment. The question for 2026 is what the second generation looks like. These modules introduced foundational concepts, but the AI tools students use today are far more capable and far more woven into their daily routines than anything the 2022-23 curricula anticipated.
Bilbao-Eraña and Arroyo-Sagasta (2025) showed that even brief, structured AI literacy training shifts teacher attitudes and awareness significantly. The Day of AI data adds a critical piece: it shows the same kind of shift happening at global scale, across 136 countries, with freely available materials that required minimal infrastructure. That’s proof of concept, and a strong one. The next step is building curricula that match the speed at which AI itself is evolving, because the students who went through these modules in 2023 are now using tools those curricula couldn’t have imagined. The foundation was laid well. The construction above it still needs to happen.
References
- Bilbao-Eraña, A., & Arroyo-Sagasta, A. (2025). Fostering AI literacy in pre-service teachers: Impact of a training intervention on awareness, attitude and trust in AI. Frontiers in Education, 10, 1668078. https://doi.org/10.3389/feduc.2025.1668078 // https://medkharbach.com/why-ai-literacy-for-teachers-goes-far-beyond-knowing-how-to-use-chatgpt/
- Chee, H., Ahn, S., & Lee, J. (2025). A competency framework for AI literacy: Variations by different learner groups and an implied learning pathway. British Journal of Educational Technology, 56, 2146-2182. https://doi.org/10.1111/bjet.13556
- Hollands, F., & Breazeal, C. (2024). Establishing AI literacy before adopting AI. The Science Teacher, 91(2), 35-42. https://doi.org/10.1080/00368555.2024.2308316
- Su, J., Ng, D. T. K., & Chu, S. K. W. (2023). Artificial intelligence (AI) literacy in early childhood education: The challenges and opportunities. Computers and Education: Artificial Intelligence, 4, 100124. https://doi.org/10.1016/j.caeai.2023.100124 //
- UNESCO. (2024). AI competency framework for students. United Nations Educational, Scientific and Cultural Organization. https://doi.org/10.54675/JKJB9835