Incorporating AI in Assignments

Levels of AI Integration

There is no "one right way" to incorporate AI into your assignments. Use your judgement to determine which level is best aligned with the learning objectives and outcomes of your assignment.

Level 1: AI Prohibited

AI does not contribute to the core competency being measured.

Ideal Assignments

In-class assessments, oral exams, analog drafting, personal reflection.

Level 2: AI Assisted

AI can be a tutor or brainstorming partner, but the final product must be human.

Ideal Assignments

Outlining, debate prep, basic coding syntax checking, grammar review.

Level 3: AI Co-Creation

AI is an industry-standard tool; students must learn to collaborate with it.

Ideal Assignments

Iterative drafting, pair programming, data visualization, prompt engineering.

Level 4: AI-Centric

Students act as editors and critics of AI-generated content.

Ideal Assignments

AI fact-checking, bias analysis, reverse-engineering AI outputs.

Integration Doesn't Mean Starting from Scratch

You don't need to throw out your favorite assignments to integrate AI. Often, the learning objectives remain exactly the same; we just need to shift how the student demonstrates their mastery. Here are some strategies to tweak your existing assignments, whether your goal is to lean into AI or design around it.

Shifting from Product to Process

Make the cognitive process visible and discourage last-minute, one-click AI generation.

Instead of grading only the final essay or report, distribute the points across the entire drafting phase.

Require students to submit their brainstorming notes, an initial outline, and a short reflection on how they revised their work. If AI is permitted (Level 2), require them to submit their prompt history as an appendix to show how they directed the AI.

Flip the Script

Elevate the task to the highest levels of Bloom's Taxonomy (Evaluating and Creating) while teaching AI literacy.

Take your standard research paper prompt and feed it directly into Gemini. Give the AI-generated essay to your students.

The assignment is no longer to write the paper, but to critique it. Students must use Track Changes to correct hallucinations, improve weak arguments, add specific academic citations from your syllabus, and write a summary of the AI's blind spots.

Hyper-Localize the Content

LLMs cannot access what happened in your classroom last week or the student's personal lived experience.

Update your prompt to require connections to highly specific, recent, or personal contexts.

Instead of asking, "Analyze the impact of supply chain disruptions on retail," ask, "Analyze the impact of supply chain disruptions on retail, referencing our guest speaker's comments from Tuesday and applying it to the local businesses in our university's surrounding neighborhood."

Change the Modality

Shifting away from text-only deliverables requires students to synthesize information in new ways.

Keep the prompt, but change the final deliverable format.

Instead of a 5-page paper, ask students to submit a 5-minute podcast (which they can script with the help of AI), participate in an in-class debate, create a digital portfolio, or sit for a 10-minute oral defense of their research.

AI Tools

Why are these the recommened tools?

SJU is a Google Workspace institution, meaning you have access to a powerful suite of integrated AI tools. Using your university account ensures a higher standard of data privacy and security compared to public, consumer-facing AI models. Below is a guide to the approved tools available to you and your students, along with ideas for how to integrate them into your courses.

Gemini

Google’s primary conversational AI model. It can process text, code, images, and audio to answer questions, brainstorm, and generate content.

Faculty Use Case (Course Design): Brainstorming assignment prompts, translating complex medical jargon into plain language for case studies, or generating diverse patient profiles for simulation labs.
Student Use Case (Assignments): Acting as a debate sparring partner (Level 2 Integration) or generating an initial draft of an essay that the student must aggressively edit and critique (Level 4 Integration).
- Gemini is not currently functional in SJU student accounts

Gemini Notebook

An AI-powered personalized research assistant that is "grounded" in your specific documents. Unlike standard Gemini, which searches the whole internet, Gemini Notebook only pulls information from the PDFs, Google Docs, Slides, and URLs that you upload to it.

Faculty Use Case (Course Design): Uploading all of your course readings for the semester to instantly generate custom study guides, FAQs, or multiple-choice knowledge checks based strictly on your syllabus.
Student Use Case (Assignments): Students can upload their own research papers and use the AI to synthesize connections between multiple journal articles, or use the "Audio Overview" feature to turn their readings into an AI-generated conversational podcast to listen to on the go.

Google Vids

An AI-powered video creation app built directly into Google Workspace. It acts as a video, writing, and production assistant all in one, generating storyboards, scripts, voiceovers, and stock footage based on a simple text prompt.

Faculty Use Case (Course Design): Quickly creating engaging, 2-minute micro-learning videos, course introduction modules, or weekly recap announcements without needing video editing skills.
Student Use Case (Assignments): Moving away from traditional essays by having students use Vids to create a persuasive public health PSA, a patient education video, or a final capstone presentation.
- Google Vids is not currently functional in SJU student accounts

Gemini in Google Workspace

The AI assistant embedded directly into the tools you already use every day. It appears as a sidebar or a "Help me write/organize" button within the document.

Gemini in Docs: Instantly draft grading rubrics, rewrite assignment instructions for clarity, or summarize long meeting notes.
Gemini in Sheets: Ask the AI to organize raw survey data, generate complex formulas, or create a project management tracker for student group work.
Gemini in Slides: Generate presentation outlines, create custom placeholder graphics for your lectures, or summarize long presentation decks into key takeaways.

Holding Learners Accountable

No matter how you decide to incorporate AI into your assignments, it's critical to the learning environment that learners abide by the standards that you have set for your course and individual assignments. Instructors should reference the SJU Policy Library for the most up-to-date policies about academic honesty, Information Technology, and more.

Field-Specific Examples

Nursing: The Flawed Care Plan (Level 4: AI-Centric Critique)

Shifting the student from rote memorization to high-level clinical judgement and auditing.

The Old Assignment: Write a standard care plan for a patient with a COPD exacerbation.
The Redesign: The instructor provides an AI-generated care plan that intentionally (or naturally) contains subtle clinical errors and generalized assumptions.
The Student Task: Students must act as the "Attending." They must use Track Changes to correct clinical inaccuracies, flag contraindicated interventions, and add specific, evidence-based rationales from their current textbook.

Occupational Therapy: The Health Literacy Translator (Level 2: AI-Assisted Process)

Mirroring a real-world clinical efficiency skill (using tools to draft patient comms) while holding the student accountable for the final clinical accuracy.

The Old Assignment: Create a patient education flyer based on a recent peer-reviewed journal article about stroke rehabilitation.
The Redesign: Students use Gemini to translate the dense, jargon-heavy journal article into a 6th-grade reading level.
The Student Task: The AI does the initial translation, but the student must verify that no clinical nuance was lost in the simplification. The student designs the final brochure and submits a "Methods Appendix" showing the prompts they used to get the AI to the right tone.

Respiratory Therapy: The SBAR Verbal Defense (Level 1: AI Prohibited / Modality Shift)

Changing the modality to a verbal, time-constrained environment completely bypasses the ability to rely on generative AI, accurately simulating the pressure of a hospital floor.

The Old Assignment: Write a 2-page case study analysis recommending a change in ventilator settings.
The Redesign: The written paper is removed entirely to measure unassisted, on-the-spot clinical reasoning.
The Student Task: Students review a patient chart and have exactly 3 minutes to deliver a verbal SBAR (Situation, Background, Assessment, Recommendation) handoff to the instructor, who is acting as the attending physician.

Health Information Management: Policy Co-Creation (Level 3: AI Co-Creation)

Teaching "prompt engineering" as a core professional competency for administrative and management roles.

The Old Assignment: Draft a standard clinic policy for telehealth data privacy.
The Redesign: Students are required to use AI to generate the foundational boilerplate for the policy, treating the AI as a legal drafting assistant.
The Student Task: Students must iteratively prompt the AI to refine the document to comply with specific state laws and the university clinic's specific software stack. They submit the final policy along with a "Prompt History" document demonstrating how they guided the AI.

Physical Therapy: Hyper-Localized Clinical Reflection (Strategy: Hyper-Localization)

AI cannot access the student's lived experience or the specific, messy realities of a real patient. This ensures the reflection is deeply personal and authentic.

The Old Assignment: Write a reflection on the standard rehabilitation timeline for a total knee replacement.
The Redesign: The prompt is anchored directly to the student's in-person clinical practicum.
The Student Task: "Ask Gemini to generate a standard 6-week rehab timeline for a total knee replacement. Then, write a 500-word reflection comparing this 'textbook' AI timeline to the actual progression, barriers, and environmental factors of a specific (de-identified) patient you worked with at your clinical site last week."

This page was created in collaboration with Google Gemini.

Page updated

Report abuse