AI tools change how many students approach academic work. Lasell's academic integrity policy states that all submitted work must be original, i.e., students can only use AI tools with instructor permission. Faculty can design AI-assisted assignments where appropriate. However, students should still cite these using standard academic practices.

Some best practices for managing student use of Generative AI are below:

Understanding AI Tools

Many AI tools exist, so you shouldn't assume students are just using ChatGPT. Since Generative AI predicts probable next words, images, and more, students can use it in various ways:

Generative Writing Tools: Tools that generate text, like ChatGPT, Claude, and Gemini.
Support Tools: Tools that assist with readability, grammar, and translation, such as Grammarly, Hemingway App, or translation services (tools like ChatGPT can do this too, but support tools exit that primarily focus on these skills).
Integrated Tools: Tools that help synthesize content and help users analyze it. For instance, NotebookLM for study guides or AI-generated podcasts.

Any of these can produce AI-generated text. So, "detecting" AI use isn’t as simple as asking students whether they used ChatGPT. Students can engage with AI in diverse, complex ways.

Identifying AI-Generated Writing

Lasell uses Turnitin's AI detector, which is helpful but not foolproof. To use it, create a Turnitin assignment and review the AI detection report after students submit. However, you should always use this report in the context of your situation, and use it carefully. The types of tools available can reflect different levels of student reliance (or over-reliance) on AI. Here are some general rules of thumb:

100% Score in Turnitin AI detector: Generally suggests copy-pasted AI content. Students may have copy/pasted the assignment prompt straight into an LLM. However, this score could also reflect original work that the student asked an LLM to proofread for grammar correctness, or it could reflect translated work. Confirm through student discussion.
Low Scores: May indicate paraphrased or lightly edited AI-generated work. Patchwork writing. So, the detector might identify entire paragraphs that may be AI while others seem likely the work of the student. The way students work with these tools is relevant here. How much is the bot just prompting them to think? How much is it "thinking" for them?

Common Characteristics of AI-Generated Text:

Generic analysis with surface-level insights. Good prompts can get better. However, unsophisticated use will miss out on specific examples that fit certain contexts.
Lack of specific examples or connections to class discussions.
Overuse of transitions (e.g., “Let's delve into…”) or excessive, fragmented subheadings. Subheadings are generally good practice, but frequent small sections may indicate AI assistance.

Best Practices for Faculty

Clarify Expectations:

Clearly state AI policies in syllabi and assignment instructions.
Provide examples of acceptable and unacceptable AI use.
Work with students to analyze AI-generated output. Ask: Is it robust? Where does it fall short in the context of your discipline?

Design AI-Resistant Assignments:

Implement multi-stage processes (drafts, peer reviews) to scaffold learners and reduce overreliance on AI.
Require personal reflection or references to class discussions.
Include in-class writing components.

Foster Ethical Use Discussions:

Discuss responsible use of AI tools.
Emphasize learning integrity and the value of original thought.
Remind students that AI can make mistakes, misunderstand context, and misrepresent complex ideas—skills they still need to master themselves.

Responding to Suspected AI Use

Start with an Open Discussion:
- Avoid accusations. Ask: "I'd like to understand how you approached this assignment."
Guiding Questions:
- "What steps did you take to complete this assignment?"
- "Did you use any digital tools for assistance?"
Review Evidence Together:
- Share the Turnitin AI report if appropriate.
- Highlight specific concerns, such as generic phrasing or lack of detail.
Determine Consequences:
- Minor Issues: Allow resubmission with clearer guidelines.
- Significant or Repeated Issues: Follow academic integrity protocols.

Grading Considerations:

Engaged Student, Minor AI Use: Address the issue while recognizing demonstrated learning.
Minimal Engagement, Significant AI Use: May warrant stricter academic consequences.

Conclusion

Navigating AI-assisted work requires a balance of clear policies, proactive teaching, and thoughtful assessment. We recommend you foster open dialogue, design thoughtful assignments, and use detection tools carefully. As AI tools evolve, continuous reflection and adaptation will be key.

For further support, contact TLC@lasell.edu or 617.243.2001.