Skip to main content
CAIL Sandbox Docs

Student Onboarding

Introduce the CAIL Sandbox by explaining what it does, its limitations, and how students are expected to use it in your course. Adapt the first-week plan below to your discipline and assignments.

Arrange Course Access

Get course approval before setting up the class or inviting students. An individual account does not enroll your course.

  1. Submit a course access request. Select A class I’m teaching or organizing, sign in with CUNY Login, and provide the requested course details. One application covers your class. Students do not submit individual access requests; they join through the course invitation link.
  2. After approval, open your course in My classes, review its enrollment dates, and create or copy its reusable invitation link.
  3. Open the Sandbox with CUNY Login. Confirm access to your course’s automatically created private group and text channel, and use the group to share models and knowledge bases.
  4. Share the invitation and follow the first-login steps below with students. Enrollment and Sandbox sign-in add them to the course group and text channel automatically.

Before the first Sandbox activity, check the roster in My classes and resolve enrollment or login problems. For help, contact ailab@gc.cuny.edu.

Before the First Class

Prepare Your Environment

  1. Email the CUNY AI Lab team to request Workspace access. Once enabled, create the models and knowledge bases students will use (see the Design & Test section).
  2. In Access, keep resources Private and use Add Access to grant your course group Read access. Apply these permissions to the model and any knowledge bases, tools, or skills you attach.
  3. Test each model with the kinds of questions your students will ask
  4. Write a brief AI use policy for your syllabus (see the template below)

Draft Your AI Use Policy

Students arrive with different experiences and assumptions about AI. Some may regard any use as cheating; others may expect it to write their papers. State what the course permits so students can judge their use against the assignment’s requirements.

Your policy should address

  • What is allowed. Which assignments permit AI use? Which do not?
  • How to cite AI use. What disclosure is expected when students use AI in their work?
  • What counts as the student’s work. Where is the line between using AI as a tool and submitting AI output as your own?
  • Privacy. What should students avoid sharing with the model? (Personal data, other students’ work, confidential materials, etc.)

Template

This course uses the CAIL Sandbox for specific assignments marked [AI-Assisted]. For these assignments, you may use the course model to brainstorm, get feedback, or explore ideas. You must disclose AI use in your submission and describe how you used it. All final writing, analysis, and argumentation must be your own. Do not paste other students’ work or personal information into any AI tool. For assignments not marked [AI-Assisted], AI use is not permitted.

Adapt the template to the uses you permit in each assignment.

Week 1. Orientation Session

Day 1. What Is This and Why Are We Using It (20 minutes)

Frame the Sandbox in your course context. Students need to understand why you chose this tool for this course.

Discussion Prompts

  • “How many of you have used ChatGPT or a similar AI tool? What did you use it for?”
  • “What do you think AI tools are good at? What are they bad at?”
  • “Why might a [discipline] course use AI tools differently than a [other discipline] course?”

Day 1. First Login (10 minutes)

Walk students through the login process. Project your screen or share a link to the Getting Started page.

  1. Open the course invitation link provided by your instructor. Going directly to the Sandbox does not enroll you in the course.
  2. Sign in with CUNY Login and complete the confirmation prompts to join the course.
  3. Navigate to chat.ailab.gc.cuny.edu.
  4. Select Continue with CUNY Login and authenticate again if prompted.
  5. Confirm that the course’s private text channel is available, then locate the course model in the selector on the right inside the message box. Course group and channel membership are added automatically through this onboarding process.

If CUNY Login stalls, try a private browser window or a different browser. If you still cannot enroll or see the course resources, contact ailab@gc.cuny.edu before continuing.

Day 1. Guided First Chat (15 minutes)

Give students a low-stakes prompt to try. This builds comfort with the interface.

Starter Prompt. “Ask the course model to explain one concept from last week’s reading that you found confusing. Then ask a follow-up question about its response.”

Circulate and answer questions while students explore the interface at their own pace.

Day 2. Critical Evaluation Exercise (30 minutes)

Once students are comfortable with the interface, introduce critical engagement.

  1. Ask students to prompt the model with a factual question from the course
  2. Copy the model’s response
  3. Identify three specific claims in the response
  4. Verify each claim using course readings or library databases
  5. Report back. Which claims checked out? Which did not? Which could you not verify?

Why this exercise. Students practice checking AI output against course readings and library sources, establishing a process they can use in later assignments.

Tip. Save 5 minutes at the end for a whole-class debrief. Ask, “What surprised you?” Compare the claims students could verify with those they could not.

Common Student Questions

“Is this cheating?” Follow the course’s AI use policy and the instructions for each assignment. Use the model only where AI assistance is permitted, disclose that use as directed, and do not submit AI output as your own work.

“Can a model access my other chats?” The model uses the current chat’s message history. If Memory is enabled or you attach another chat as context, information from outside the thread can also be included.

“Who can access my saved chats?” Administrators can access stored chats and system logs. Other students see content you share with them or post in a course channel. Requests to AI providers are configured for zero retention; use of prompts and responses for model training is prohibited. Saved Sandbox history is separate. See Privacy and Chat History.

“What if the model gives wrong information?” AI models can generate plausible but inaccurate text. Check their claims against course readings and other sources before using them in your work.

After Week 1

Build on the Foundation

  • Reference the onboarding experience in later assignments.
  • Invite the class to revisit and revise the AI use policy mid-semester as students’ understanding of AI evolves with use.
  • Share interesting student discoveries with the class. When someone finds a creative use or catches an interesting error, make it a teaching moment.

Monitor and Adjust

  • Check in with students about their AI use every few weeks; ask what is working and what is frustrating
  • Adjust model configurations based on how students actually use them
  • Update your knowledge bases as the course progresses (add new readings, remove outdated materials)

Callout

For TAs and teaching staff. Spend 30 minutes exploring the course model before the semester starts so you can answer students’ questions from experience.

Additional Resources

← Return to Sample Activities | Continue to Use Cases →