Microsoft has put out a set of AI resources for American teachers, produced with Cyberlite and aimed squarely at the people who write curricula and are trying to fit the technology into lesson plans. The centerpiece is a video in which Cyberlite's Nina Bual walks through using generative AI for lesson outlines, quizzes and the rest of a teacher's daily paperwork, alongside exercises teachers can run on their own before trying anything in a classroom. The package is less about AI in the abstract than about one skill, prompting, and the limits Bual argues should come attached to it.
Bual's framing for that skill is a comparison with ordinary internet search. Search is simple by nature, she says, and it is now being displaced by a different mode of interaction: writing prompts. Search has limits; prompts work differently. A substantial share of the video goes to why this matters in 2026 and how a teacher builds the habit, and the Microsoft materials include exercises that push teachers to try the approaches themselves and then carry them into the classroom.
She also pushes storytelling as the way in, on the argument that it helps people see how these tools get used beyond the classroom, in the working lives of people who use them professionally. And she recommends doing this with colleagues and administrators rather than alone: when people get excited about a new idea, they tend to turn to the people around them to talk it through and compare notes.
The closing section is a list of constraints. Bual tells teachers to account for the environment they are working in. Schools in different countries pick different tools, and generative AI moves very fast, so teachers should check which tools their school and district actually use, and read the rules already in force, before bringing AI into a lesson.
Then she turns to attention and screen hygiene. Adopting generative AI, in her account, means setting boundaries, capping screen time, and keeping up relationships with family and friends so that dependence on the technology does not grow past the useful point. Meet those conditions, she argues, and teachers can put AI, generative and otherwise, to work across a wide range of tasks.
There is already evidence of what that looks like day to day. Ana Sepulveda, a Dallas teacher quoted by Colorado Sentinel, says AI helps her translate materials for students, plan lessons, communicate with parents and raise student engagement. Those are ordinary uses of a general-purpose tool, and in some narrow tasks it can come close to instructing children on its own, which is why the material points to Amanda Sturgill of the Center for Engaged Learning at Elon University and her argument for keeping a person in the loop.
What strikes me about the package is the choice of subject. Defining AI literacy for teachers as prompting skill is a convenient definition for the company shipping it: prompting is a skill you practice inside a product, it improves with use, and it produces the feeling of competence quickly. The harder questions in a school are elsewhere, and this set of materials, as described, is quiet on nearly all of them. Nothing here on grading, on what a teacher does when the model is confidently wrong in a subject the teacher does not know cold, on what students are doing with the same tools, or on what happens to student information typed into a prompt box. The most valuable advice in the video is not a technique at all; it is Bual's instruction to find out what your district permits before you start.
That advice has an order of operations built into it, and it is the right one: the policy first, the prompting second. The problem is that the training arrives on a teacher's screen on its own schedule. Where a district has not yet decided anything, a teacher who finishes the video is trained, equipped and ready, and whatever she chooses to do next becomes the district's de facto AI policy for one classroom.