Control Alt Achieve: Did a Robot Write this Report? Managing AI Cheating

Generative AI is a powerful tool that can be used to support teachers and students. Most of the time when I am sharing about AI or doing AI trainings, I often focus on the positives aspects of this technology and how it can be used to benefit education.

Unfortunately, just as AI can be used to generate lesson plans, provide helpful feedback, and serve as a personalized tutor, it can also be used to write a paper, provide answers, and do students' work.

In short, AI can be used to cheat. Depending on the study you look at, research shows that between 60% and 70% of students admit to using AI to cheat.

It is critical for schools to explore and address this concern so that students can benefit from the support of AI, but not abuse this tool and ultimately hurt their own learning.

But how can we manage this? Recently I recorded a 50-minute video where I do a deep dive into addressing AI cheating. To be clear however, there are no easy answers. We are all learning this together and my presentation simply shares hopefully helpful ideas. I welcome your suggestions, thoughts, questions, comments, and opinions!

See below to watch the recorded video and access all of the support materials.

▶️ Recorded Training (50 minutes)

📜 To get a PD Certificate watch the video through EdPuzzle

👀 If you do not want a certificate watch the video on YouTube

🧰 Session Resources

Session Resource Document - bit.ly/curts-aicheat
Session Slideshow - View slides - Copy slides

📔 Define Cheating

The first step to help manage cheating with AI is simply to define what it means to cheat with AI. There are many ways that AI can be used during a school project, with some uses being beneficial and some being cheating. This is something that each teacher needs to explore and decide what is appropriate in their class, and perhaps per assignment.

As one example, in your own opinion consider which cases below are acceptable uses of AI when writing a paper:

Using AI to brainstorm ideas for the paper
Using AI for spell check and grammar assistance
Using AI to check for plagiarism
Using AI to write an outline for the paper
Using AI to provide feedback on how to improve the paper
Using AI to paraphrase portions of your writing
Using AI to summarize content being read for the paper
Using AI to answer questions about the content
Using AI to write a rough draft of the paper
Using AI to find informational sources for the paper
Using AI for feedback on style and tone
Using AI to properly cite and reference sources

There are no "correct" answers to that question. This is something you need to wrestle with and share your expectations with your students.

✅ Adjust Assignments and Assessments

The next thing to consider is the potential need to modify your existing assignments to make them more AI resistant. Knowing that students have access to AI, and that many may use it for cheating, what are some ways you have adjusted your assignments to help reduce that chance? Here are a few thoughts:

When creating assignments:

AI Test the Assignment - Play the role of a student trying to cheat with AI on your specific assignment. This can help show:

How susceptible your assignment is to AI and if you will need to make some adjustments.
What a typical AI response to your assignment looks like so you can better spot AI uses from students.

Multi-Stage Assignments - Break assignments into multiple stages, requiring drafts, outlines, or brainstorming sessions, making it difficult for students to solely rely on AI.
Personalized Prompts - Craft assignments that require personal reflections, unique perspectives, or specific experiences, making it harder for AI to generate relevant responses.
In-Person Work - By supervising writing and other work in real-time, teachers can be sure the student is doing the work. This can also provide good samples of student work to compare future work against.

Another option is to include more oral responses for assignments.

Interactive Whiteboarding - Using a digital or physical whiteboard, students can show their thought process as they work through a problem.
Oral Examinations - These require real-time responses from the student and also check the student's ability to articulate thoughts verbally.
Performances - In drama, music, or language classes, live performances showcase a student's practical proficiency.
Socratic Seminars - In this form of assessment, students engage in a deep discussion about a topic, showing their understanding, interpretation, and analysis.
Debate or Speech - Real-time dialogue or presentations require a deep understanding of the topic, clear articulation of thoughts, and quick thinking.
Classroom Discussion - Students' comprehension can be gauged by their ability to engage in discussions, ask insightful questions, and provide meaningful contributions to the topic at hand.
Use tools that can assist in evaluating oral responses such as Snorkl - snorkl.app - Blog post - Demo video (12 minutes) - Demo class to test as student

Beyond that, other ideas for alternate assignments and assessments can include:

Problem-Based Learning - This approach presents students with practical, real-world problems that they have to solve, often collaboratively.
Practical Experiments - In science or engineering classes, physical experiments that demand hands-on knowledge can be valuable.
Group Projects - While it's possible for one person in a group to cheat, it's less likely if the group must collaborate and discuss their work in person.
Peer Teaching - A student who has fully understood a concept can often teach it to others.
Internships or Field Work - Real-world experience in the form of internships, co-op programs, or field work is also a great way of assessment.
Community-based Projects - By involving students in projects that benefit their community, educators can assess not only their academic knowledge but also their ability to apply it for social good.
Game-Based Learning - Educational games that require problem-solving can be a good form of assessment.

📑 Encourage Proper Citation

Another idea to help address cheating is to encourage students to acknowledge how they use AI for assistance in their assignments. This transparency can help highlight the appropriate use of AI as an assistant.

For traditional citations of AI generated content students can use:

MLA format - https://style.mla.org/citing-generative-ai/
APA format - https://apastyle.apa.org/blog/how-to-cite-chatgpt

For general citation of AI generated content:

For any assignment on which a student received any form of AI assistance, have them include an "AI Credits" section to the bottom of their work.
Here they can simply disclose how they used AI for this assessment including brainstorming, outlining, feedback on their writing, and such.
They can simply copy and paste the responses from the AI if that is most convenient.
Example - Google Docs link

🔎 Monitor AI Usage

Even with clear expectations and adjusted assignments, you will still need to be on the lookout for content that might be AI generated. The more you know your students, the easier it will be to spot potential cheating. Keep an eye out for signs of misuse, such as suspiciously high-quality work, work that does not match the student's usual writing style, or students submitting similar work.

There are many tools designed to analyze text to determine the likelihood that it was generated by AI or a human. However these "AI detector" tools are not 100% accurate, so only use them to facilitate conversations about misuse concerns.

Example - GPTZero - gptzero.me

This tool even includes a warning that the results are not guaranteed accurate:

"This result should not be used to directly punish students. For a more holistic assessment and responsible use of GPTZero results, read our five steps towards responsible AI detection."

So why are AI detections tools not reliable? There are many reasons:

Not 100% accurate (55% to 97%)
Biased against language learners
Biased against autistic students
Susceptible to paraphrasing
Susceptible to more advanced AI generators
Susceptible to more advanced prompts
Susceptible to "humanizing" tools

Using AI detectors is like trying to unbake a cake. You are starting with the finished product and then trying to go backwards, using algorithms, to try to predict how the final product was created.

Instead we need to use tools and techniques that allow us to monitor the student's work as it is being created, following along with them as they work through the process to get to their final product.

Version History

The "Version History" tool in Google products can be used to show every change made to a document, who made the change, and when the change was made.
This can help show if work was written by the student or pasted in from another source.

Revision History

The Revision History extension - Chrome Web Store link - can show:
How many edits students made to their essays
How long they spent writing
What snippets they copy/pasted
A video replaying the document being written

Brisk Teaching

Brisk "Inspect Writing" - briskteaching.com/inspect-writing-demo - Extension link - Blog post
Can be used to play back a video of the writing process, showing spots where pasting was done.

Other tools

If you do not use Google tools, there are other options for monitoring the writing process.
ClassCompanion - classcompanion.com - Blog post
Rumi - rumidocs.com

🧭 Set AI Guidelines

When you are ready to develop AI guidelines for your class, there are many excellent resources available online to use as templates or examples to draw from:

Policy and Guideline Frameworks and Guidance

Sample Guidance on the Use of AI from TeachAI

Overview - teachai.org/toolkit
Webpage version - Resource link
Google Docs version - Resource link

Planning Guide for AI from Michigan Virtual - Resource link
Framework for Implementing Artificial Intelligence in K-12 Education (ILO Group) - Resource link
AI Toolkit from Innovate Ohio - Resource link
North Carolina Generative AI Implementation Recommendations and Considerations for PK-13 Public Schools - Resource link
Generative Artificial Intelligence (AI) in K-12 Classrooms (Oregon DOE) - Resource link
Human-Centered AI Guidance for K-12 Public Schools (Washington OPSI) - Resource link
AI Guidance for West Virginia Schools - Resource link
Artificial Intelligence Guidance Brief (Kentucky DOE) - Resource link
Artificial Intelligence Guidance for K-12 Classrooms (Mississippi DOE) - Resource link
AI Guidance for Arizona Schools - Resource link
K-12 Generative AI Readiness Checklist - Resource link
AI Guidebook (Chicago Public Schools) - Resource link

Policy and Guidelines Examples

Syllabi Policies for AI Generative Tools - Large collection from Lance Eaton

Google Docs version - Resource link
Google Sheets version - Resource link

NEOLA examples

Pleasant Local Schools - Marion, OH - Resource link
Princeton City School District - Cincinnati, OH - Resource link

AI Syllabus Policy Statement - Dr. Torrey Trust - Resource link
University of Kentucky - Resource link
Lawrence School - Sagamore Hills, OH - Resource link
Plano ISD's Generative Artificial Intelligence Position and Guidance - Resource link
All Saints’ Episcopal School A.I. Course Policies by Subject - Resource link
Beverly Hills Unified School District Academic Honesty Policy - Resource link
Syllabus Considerations for AI - Chagrin Falls, OH - Resource link
Guidelines for AI Use in College Writing - Champion Local Schools, OH - Resource link
AI Guidelines for Teachers - Norfolk Public Schools - Resource link
WCDSB AI Guidelines for Students and Educators - Resource link
Missouri School Boards' Association - Sample Policy - Sample Use Plan
Strongsville City Schools (Ohio) - Resource link

Additionally here are some questions to explore, discuss, and answer when developing AI guidelines and policies:

What are positive ways that AI can be used for students, with students, and by students?
What are concerns and potential misuses of AI?
How does AI align to the school's educational objectives?
What are the goals for using AI in school?
How will the school define academic dishonesty when using AI tools? What does it mean to cheat with AI in different subjects, grades, etc?
What specific guidelines or rules will the school provide to students on how AI should be used?
How will the school monitor student use of AI in the classroom to ensure that students use AI tools ethically and responsibly?
How will the school address problems that may arise from the inappropriate use of AI in the classroom?
How will the school address potential biases from AI tools, and teach students to critically evaluate AI-generated output?
How can AI tools be used to support students with disabilities, while ensuring accessibility?
How can the school ensure all students have equal access to AI tools, regardless of socioeconomic background or technological gaps?
How will the school address data privacy issues and protect student data?
Who needs to be involved in the creation of these guidelines?
How will the school remain up-to-date on new AI developments and revisit these guidelines as needed?
How will the guidelines be shared with all stakeholders including students, parents, and staff?
What specific training and support will teachers receive to help them effectively manage and guide student use of AI?

You can use your answers to those questions, along with generative AI to help develop your class guidelines. Example prompt:

Below are my answers to key questions about my philosophy of AI in education. Using the content from my answers, generate a document titled "Student AI Guidelines" to be given to my class of students that provides detailed guidelines on the appropriate use of AI in our class. Organize, arrange, and summarize the content as needed to create a clear and structured set of guidelines for the students.

🎓 Educate Students about AI

To help address the potential for cheating with AI, we also need to educate our students on the ethical use of this powerful tool. We need to have discussions with students and teach them about the pros and cons of using AI tools including:

Positive ways that AI tools can assist them.
The importance of academic integrity and avoiding plagiarism and cheating.
The importance of data privacy and not sharing personal information.
How AI-generated content can be inaccurate and unreliable.
How AI-generated content can include biases

Thankfully there are many free lessons, activities, videos, and more that can be used to help teach our students about this. Here are some examples of AI lessons for students:

"Free Resources to Teach Your Students about AI" - Blog post
Experience AI - Resource link
Discover AI in Daily Life - Resource link
AI Literacy Lessons - Resource link
How AI Works - Resource link
Day of AI - Resource link
aiEDU

TeachAI - Resource link
Intro to AI - Resource link
Flex Plans - Resource link
AI Challenges - Resource link

Classroom-Ready Resources About AI For Teaching (CRAFT) - Resource link
DAILy Middle School AI Curriculum - Resource link
Ethics of AI Curriculum - Resource link
DAILy Workshop - Resource link
ISTE Hands-On AI Projects for the Classroom

Elementary Teacher - Resource link
Secondary Teachers - Resource link
Electives Teachers - Resource link
Computer Science Teachers - Resource link
Ethics and AI - Resource link

AI for Education - Resource link
TCEA AI Literacy Framework - Resource link
Teachable Machine - Resource link
Say What You See - Resource link
Odd One Out - Resource link

👩 Engage Parents and Guardians

In addition to the students, we also need to involve parents and guardians in the conversation about the responsible use of AI tools and the importance of academic integrity.

Educate them about current AI tools and how they can be used.
Make sure they are aware of the guidelines you have for your class.
This can create a supportive network that reinforces responsible behavior.

📰 Stay Informed

Finally one of the best ways to address potential AI cheating is to continue to educate yourself on AI advancements. Regularly update yourself on the latest developments in AI and education to anticipate potential challenges and adapt your teaching strategies accordingly.

Here are a few resources that I have found helpful to stay on top of AI in education:

Readings

Artificial Intelligence and the Future of Teaching and Learning from the U.S. Office of Educational Technology - Resource link
ChatGPT and Beyond: How to Handle AI in Schools from Common Sense Education - Resource link
AI and Education: Guidance for Policy-Makers from UNESCO - Resource link

Books