Students Want to use AI, Teachers Fear Cheating
I asked fellow educators to write about classroom AI. I fear that not discussing cheating first may impede appreciating those tips. Thus, I wrote these ideas for detecting AI in student work.
Over the next several weeks, teachers are collaborating to create helpful articles about AI in education for our staff newsletter. It’s usually my responsibility to write a professional learning section of the newsletter. Having guest writers on this section is new, but probably overdue. I hope they find positive support as they are stepping out of their comfort zone to write these articles. Having them step out of their comfort zone reminded me how much I was enjoying doing just that by posting to this substack. Thus, I grabbed the article I wrote for the teacher newsletter queuing up the guest writers and turned it into a fresh Substack post. Whenever granted permission, I’ll share the writing of my colleagues here as well.
All the best,
Aaron
Tips for Detecting AI in Student Work
As you may know, we will be featuring guest teacher-writers over the next several weeks. They are writing about their experience with AI in the classroom. They have a lot of helpful tips that may give teachers new superpowers. I fear that without discussing the biggest fear of AI in the classroom — cheating — it may be difficult to really dig into what our colleagues are offering. Thus, I spent some time listening and learning from other high schools teachers and wrote this list of hopefully helpful suggestions.
Oh, and as I believe in citing my AI collaborators, I used Google Bard (a new competitor to ChatGPT) to help me brainstorm and to paraphrase my oftentimes overly wordy writing.
Detecting students using AI for assignments can be tricky. False accusations of cheating (like here, here, and here) have led some universities to drop AI detection tools. That doesn’t mean that we’re powerless in helping our students make ethical decisions when using AI. Here are some tips for detecting AI in student work:
Analyze Their Work:
Content Quality: AI-generated work can be grammatically correct and well-structured, but it may lack depth, originality, and critical thinking. Look for generic arguments, overly factual language, and a lack of personal insights or experiences.
Style, Voice, and Tone: AI struggles to mimic individual writing styles. Check for inconsistencies in tone, vocabulary, and sentence structure compared to previous work. Sudden improvements in writing quality could also be a red flag.
Source Checking: Verify the citations in the work. Are they credible sources?Do they actually support the claims made? AI-generated work often relies on generic references.
If you suspect unethical AI use, ask the student directly:
Ask the student to explain their work in their own words. If they struggle to articulate concepts or provide genuine understanding, it could be a sign of AI use.
Unscripted tasks: Give the student impromptu questions or writing prompts related to the assignment. Can they adapt and think critically on the spot?
Oral presentations: Observe the student's presentation skills. Can they answer questions confidently and defend their work without relying on prepared scripts?
Open Communication: Discuss your concerns with students openly and emphasize the importance of academic integrity. When asked, most students are pretty open. They know when the jig is up.
Remember:
AI Can be a Collaborator and a Co-Pilot: AI can be a brainstorm partner, can help summarize text, and even shorten wordy writing. While collaboration should be cited, using AI does not mean
Set Expectations: Edutopia suggests using red, yellow, and green signals to help students understand when it’s ok to look to AI as a co-pilot. Some MHS teachers used the same signals in the early days of Chromebooks to help students to understand when it’s ok to have a device out.
Focus on tasks and prompts: Instead of solely concentrating on catching AI use, prioritize creating an environment that encourages authentic learning and critical thinking. The more your prompts require personal experiences, the less you can collaborate with AI.
Stay adaptable: AI technology is constantly evolving, so stay informed about new developments and adapt your detection methods accordingly.
Ultimately, the best approach is a multi-layered one that combines careful analysis and empathetic, direct engagement. It is a bit ironic that there is currently no algorithm that will clearly and competently identify the generative work of other algorithms. With these tips, it is my sincerest hope that you can create a learning environment where AI becomes a tool for learning, not a shortcut to phony fulfillment.
With that out of the way, I look forward to sharing the helpful tips from some of my favorite high school teachers in the coming weeks.
