Teach the AI is one of Packback's AI Integrated Assignment types. In it, students explain a concept to an AI that acts like a novice learner and pushes back with questions, so they have to teach clearly and accurately to demonstrate real understanding. It's especially well suited to building communication skills, and you are in full control of the topic, objectives, and grading criteria your students work from.

This article covers what Teach the AI is, how to build one, and how to grade student submissions. For how the Activity Library works, how activities and assignments relate, and how to reuse or edit activities across assignment types, see Managing AI Integrated Assignments.

• Building a Teach the AI Activity
• Previewing the Activity
• Adding It to an Assignment and Publishing
• Grading a Teach the AI Submission

Building a Teach the AI Activity

To start a new activity, go to the Activity Library, click Add, choose New Activity, then select Teach the AI. The builder is organized into the sections below, from top to bottom.

Activity Name

This defaults to a placeholder like "Teach the AI #1." Rename it to something you'll recognize later in your Library, especially if you plan to reuse it.

Topic

A free-text field for the concept students will teach. Students will explain this topic to an AI acting as a novice learner, so be specific about the concept you want them to demonstrate.

Learning Objectives

Add between 2 and 6 objectives (this is required). Type an objective into the text box and click Add. Each objective becomes its own row that you can edit, reorder, or delete. These objectives anchor the activity: they drive the concept map, guide the AI's questions, and appear to students as they plan and teach.

Additional Details (optional)

This collapsible section lets you give students and the AI more to work with:

• Key Terms: Enter one term per line in Term: Definition format. Students will see hoverable definitions for these terms while they teach.
• Source Material: Source material scopes both the student's reference panel and the questions the AI asks, keeping the conversation grounded in your course content.

Concept Map

Click Generate Concept Map to create a map from your objectives (you'll need to add objectives first). Packback generates three editable concepts, each with a title, a description, and its own list of common misconceptions. You can edit any of these, use + Add Concept to add more, or Clear to start over. The concept map is what powers the feedback students see when they review their teaching conversation, and it feeds the Concept Teaching Summary you'll use when grading, so it's worth reviewing each concept and its misconceptions for accuracy.

Rubric

Build a rubric of 2 to 6 categories, each scored on a four-level scale: Missing, Emerging, Developing, and Strong. Every cell is fully editable, and you can use + Add Category to add more or the Compact/Expand view toggle to change how much you see at once.

The rubric opens with generic default categories (Planning, Explanation Quality, Concept Mastery). These are meant to be customized. For a strong activity, replace them with categories specific to your topic, for example Scientific Accuracy, Critical Thinking, and Communication.

Currently, all assignments are scored out of 10 max point.

Previewing the Activity

The Preview Assignment section includes a Preview as Student button. This button stays disabled until you've added both Learning Objectives and a Concept Map, so complete those sections first.

Preview as Student lets you step into the activity exactly as your students will and teach the concept to the AI yourself. Use it to confirm everything is configured correctly: check that the AI's questions stay grounded in your topic and source material, that your objectives and concept map produce the kind of conversation you're expecting, and that the rubric fits the work students will actually do. If anything feels off, you can go back and make adjustments before publishing.

Beyond checking the setup, working through the activity yourself is worth doing on its own. It gives you a feel for where students are likely to get stuck and what a strong explanation looks like, which is helpful to keep in mind when it's time to grade.

Adding It to an Assignment and Publishing

At the bottom of the builder you have two options:

• Save to Activity Library saves the activity for later without scheduling it.
• Add to Assignment saves the activity and takes you to the scheduling screen.

On the scheduling screen, set:

• Assignment Name
• Start Date and Time, with time zone
• Due Date and Time, with time zone

Then choose Save as Draft to keep working, or Publish Assignment to make it available to students on the start date.

Once published, the assignment appears under the Assignments tab, where you can see its Type, Start and Due Dates, Progress ("X Completed"), and Status. The Actions menu lets you Edit or Preview the assignment.

Grading a Teach the AI Submission

Finding submissions to grade

From the Assignments tab, each assignment card shows its submission counts ("X/30 Submitted | Y Ungraded") along with Preview, Edit, and View Submissions links. Click View Submissions to open the roster table, which lists each student's Name, Section, Submitted At time, Grade, and Status. You can filter by Section or Status, and use Download CSV to export the list.

Each row shows a Grade button for ungraded work, or a clickable point score with a Finalized label for work you've already graded.

The grading screen

Clicking Grade opens a two-pane view.

The left pane shows the student's work for reference. The Overview tab displays their Reflection (the prompt and the student's written answer) and the full Conversation Transcript, including the turn counter (for example, "8/15 turns") and the scrollable exchange between the student and the AI.

The right pane is where you score the submission. At the top, student pagination (for example, "Jess Bautista: 1 of 4") lets you move between submissions with the previous and next arrows.

Concept Teaching Summary

The core of the scoring panel is the Concept Teaching Summary: one row per learning objective, each auto-tagged by the AI with a status of Not Taught, Misconception Present, Partially Explained, or Clearly Explained.

Expand any concept to see:

• The Objective text.
• An editable Status dropdown, so you can override the AI's call if you disagree.
• AI-written Notes explaining the reasoning behind the status.
• An Evidence panel quoting the exact supporting line(s) from the transcript. If there are several quotes, they're paginated (for example, "1/2"), and clicking through them highlights the exact message in the transcript on the left. Concepts marked Not Taught show "No supporting quotes found" instead of evidence.

The Concept Teaching Summary is a starting point, not the final grade. Use the AI's tags and evidence to move quickly, but rely on your own judgment and adjust any status that doesn't match what you see in the transcript.

Rubric and comment

Below the summary is your Rubric, with the categories and levels (Missing, Emerging, Developing, and Strong) you defined when building the activity. Click a level for each category to assign its points; the level's criteria text appears, and a running Total Points updates live at the top and bottom of the panel. You can also add an optional Overall Comment with feedback for the student.

Submitting and editing grades

You have two submit actions:

• Submit finalizes the grade and keeps you on that student.
• Submit & Next finalizes the grade and jumps to the next ungraded submission, skipping any that are already finalized.

Once submitted, the record becomes Finalized and the panel locks. To make changes, click Edit Evaluation to reopen it.

Have questions?

If you need technical assistance or have questions about your Teach the AI assignment, please reach out to our Support Team at help@packback.co.