Differentiating AI Literacy for Mixed-Ability Classes (Virtual & Online Schools)

An AI literacy lesson in an online school has to work for a student who has been building chatbots since middle school and a student who has never opened a model and is not going to turn a camera on to admit it. Running four separate lessons is not sustainable when you are also teaching three other courses across synchronous and asynchronous blocks. The alternative is one lesson with four exits, planned once, where the differentiation lives in the task demand rather than in the content.

One stimulus, four demands

Start every student in the same place: the same short scenario, the same output from a model, the same three-minute look. Common ground matters more online, where you cannot glance around a room to see who is lost. Then split the demand.

1. Scaffolded. Given a model output and a list of five possible problems with it, identify which two apply and mark where in the text they appear. The thinking is judgment; the language is supplied.
2. Core. Same output, no list. Students name two problems in their own words and quote the evidence.
3. Stretch. Rewrite the prompt so the output would be more reliable, then explain what change in the prompt caused what change in the response.
4. Challenge. Argue when this tool should not be used for this task at all, with a stated criterion someone else could apply to a different case.

Four levels, one scenario, one set of materials. The prep cost is writing the list for level one and the criterion prompt for level four; everything else is the same lesson.

Letting students choose, and correcting the choice quietly

Publish all four and let students pick. Most self-select accurately when the levels are described by what you do rather than by how hard they are. The ones who pick wrongly are visible in the submissions within a week, and a private message reassigning a student in either direction lands better online than a public grouping ever does. Set a rule that anyone finishing early moves up a level rather than waiting, which converts spare time into work instead of into a chat window.

Having a bank of ready scenarios is what makes this repeatable rather than a one-off. The task sets in AI Literacy & Cybersecurity Activities | Artificial Intelligence, Prompt Engineering & Online Safety | Computer Science Bundle | Grades 7–10 cover prompt work and online safety scenarios you can drop straight into the stimulus slot, and because the files are editable you can strip a scenario down for the scaffolded exit and extend the same one for challenge.

Keeping asynchronous work honest

The obvious problem with AI literacy homework is that a model can complete most of it. Design around that rather than policing it. Tasks that resist automation share a shape: they require the student's own earlier work, a personal example, or a judgment about a specific classmate's answer. Ask students to critique the model output they generated themselves and paste both. Ask them to compare a model's answer to a claim made in the previous live session. A response that references what happened in your class on Tuesday is a response a student wrote.

Discussion-shaped material gives you plenty of those prompts. The scenarios in Digital Citizenship Activities | Cyberbullying, AI, Social Media & Media Literacy Lessons | Grades 7–10 are built around situations students have opinions about, which is what makes a written response worth reading in an asynchronous forum where a generic answer is easy to spot.

Assessing four levels without four rubrics

Use one rubric with a single column for the quality of reasoning and a note of which level the student attempted. A strong level two answer and a weak level four answer are different pieces of information, and you want both in the record. Over a unit, track movement between levels rather than the score itself; that movement is the differentiation working. For units where students need the wider computing background to reach the challenge level, a broader collection such as the Computer Science Activities Bundle | Artificial Intelligence, Python Programming, Cybersecurity & Digital Literacy | Middle School Grades 7–10 gives you the neighboring topics to pull from when a student is ready to connect prompt behavior to how the systems underneath actually work. Done well, the live block sounds like one conversation with four depths in it, and nobody has to announce which one they are in.

Related reading from the Teacher Hub

• Meaningful Homework and Independent Practice for Programming Basics (Virtual & Online Schools)
• A No-Prep Sub Plan for Creative Writing (Student Teachers)
• Using Digital Tools to Teach Narrative Writing (Student Teachers)