Toolkit and Resources: AI in Education

Welcome to the CTL’s Comprehensive Toolkit on AI in Education

This guide serves as your primer on integrating Artificial Intelligence (AI) and generative AI technologies, like ChatGPT, into the academic environment at Moraine Valley Community College (MVCC). Our aim is to help you to navigate these innovative tools within the framework of academic integrity, providing both insights and practical applications.

Artificial Intelligence – MVCC Policy Statement

Crafting Instructor/Course Specific Syllabus Statements

Exploring AI Detection Tools and Identifying Red Flags

Current State of AI detection

Steps to Take If You Suspect Unauthorized A.I. Use

Using AI Generative Technology

Ethical Considerations

 

 

Artificial Intelligence MVCC Policy Statement

Below you can find the MVCC college-wide statement on generative AI and ChatGPT use in the classroom. Please include this in your syllabi.  

“All students are expected to maintain academic integrity in their academic work and honesty in all dealings with the College. A student who cheats, plagiarizes, or furnishes false, misleading information to the College violates the academic integrity code of conduct.

With an increase in the availability of online writing tools, students should be aware that submitting work created by AI software could be considered plagiarism unless an instructor has pre-approved its use.”

We also suggest you consider crafting a course specific AI policy that gives students guidelines that are more specific to your course. We share some resources on this in the next section.

 

Crafting Instructor/Course Specific Syllabus Statements

The inclusion of AI technologies in coursework prompts a need for clear guidelines within your syllabi. A discussion of generative AI in your syllabus and in the classroom can help students to better understand these tools and can also help to prevent misuse. You will generally want to consider whether you would like to use a strictly prohibitive statement, a flexible/partial use statement or a wider use statement. As you consider your policy, you may want to use the resources below.

 

Exploring AI Detection Tools and Identifying Red Flags

The current landscape for AI detection tools is rapidly evolving, yet no foolproof system exists. Educators are encouraged to view these tools as part of a broader dialogue about academic integrity rather than definitive solutions. Many tools are currently available to detect AI. None of them are completely accurate, but the three below do fairly well in differentiating AI and human writing.

  • GPTZero: A tool designed for educators to detect potential AI-generated text. This tool was trained on student writing and is meant for educational settings.
  • ZeroGPT: Offers more extensive analysis capabilities in its free version.
  • Copyleaks: A comprehensive plagiarism detection service with AI-specific features. A study from Cornell found it beat GPTZero, but you do need to pay for an account. 
  • Other detectors available: This is an article covering the “12 best AI detectors for 2024” Please note that different sites promote different AI detection tools. This is just one example.

Given that these tools are not completely reliable or valid, it is advisable to use more than one AI detector when you have concerns with a piece of student writing. It can also be helpful to look for red flags. Below are some common red flags* and then a few links to sites that discuss this in more depth.

  • Using complex phrases and terms that a typical student would not use.
  • Writing that does not match the directions.
  • AI language generators often guess at sources it is not familiar with and say things like “The book “Prophet Song” probably covers . . .”.
  • Perfect grammar and spelling
  • Vague or generic – Can lack emotional quality or tone (lack of personal touch)
  • Unnecessary verbosity and extra transitions
  • Uniformity in sentence length and structure – repetitive content
  • Includes made-up sources
  • AI writing detection: Red Flags (Montclair State University)
  • Unmasking AI: A guide to detecting artificially generated content (AIContentfy)
  • Professor guide to AI detection (Packback)

*It is worth always keeping in mind the words of AI in higher education expert Ethan Mollick, “You might think you are good at detecting AI writing, but you are just okay at detecting bad AI writing, and you combine that with your own biases and heuristics about who might be using AI.“

  • Both parts of this statement are so important. First, ‘you are just okay at detecting bad AI writing.’ That means that we are not going to be able to detect skilled use of AI for writing. If students want their AI writing to be undetectable there are many free tools out there that can do that for them. Second, ‘you combine that with your own biases and heuristics about who might be using AI.’ This refers to the facts that we all have biases and assumptions about different kinds of students and who is and is not likely to cheat. These biases and heuristics have varying degrees of accuracy, but even the most accurate heuristic is wrong some of the time.
  • Remember that a positive result from an AI detector and a few red flags are NOT definitive judgments of academic misconduct. They are only indicators that the writing requires further scrutiny and human judgment from the faculty, along with the application of their specific academic policies and expectations. Positives can be challenged by students who can provide evidence of their originality and writing process, such as drafts, notes, sources, etc. Faculty should be open to having non-accusatory conversations with students and listen to their explanations. Remember that you will be wrong at least some of the time and when you inaccurately accuse an innocent student of cheating, you may contribute to some serious and negative life-altering consequences.

 

Current State of AI Detection:

There are a number of challenges in using AI detection tools. A few of these challenges include: 

  1. Complexity of AI-generated Content: The ability of AI to mimic human writing styles closely makes it difficult for detection tools to accurately distinguish between human and AI-generated texts.
  2. Bias: Detection algorithms may disproportionately flag content from neurodivergent individuals or non-native English speakers as AI-generated, introducing bias.
  3. Evasion Techniques: As users and developers of AI writing tools devise methods to bypass detection, maintaining effective detection capabilities becomes increasingly challenging.
  4. Reliability: The consistency of AI detection tools is a concern, with some tools providing contradictory assessments for the same content.
  5. Grammar-based tools: Grammar tools such as Grammarly can reword phrases and sentences using AI. This rephrasing commonly triggers AI detection. You need to decide whether these tools are acceptable to use in your classes. 

The accuracy of AI detection tools varies, with most achieving moderate success at best. Their ability to identify AI-generated content is often outpaced by the advancements in AI technology, leading to a continuous cat-and-mouse game between generation and detection capabilities.

The occurrence of false positives, where human-generated content is misidentified as AI-generated, can have severe consequences. It risks unjust accusations of academic dishonesty, potentially damaging students’ reputations and trust in the educational system. Educators are strongly encouraged to use AI detection tools judiciously, emphasizing dialogue and additional verification methods over sole reliance on these tools.

Remember that false positives are more likely when:

  • the writing sample is shorter
  • the writing follows a stylized pattern (like a list)
  • the writer has a more rigid communication style
  • the writer uses an AI-based tool such as Grammarly or Quillbot to help refine their writing

Here are a few resources to explore on the effectiveness and use of AI detectors:

 

Steps to Take If You Suspect Unauthorized A.I. Use

When you suspect a student may have inappropriately used AI, we suggest that you do the following:

  • Initial suspicions and observations. You need to start with familiarizing yourself with the student’s typical writing style and patterns. Anomalies in writing style or unexpected depths of knowledge in submitted work can be initial indicators of AI use. You can also check the red flags that are discussed elsewhere in this toolkit. 
  • Check your biases. On what are you basing your belief of dishonesty? If another student had turned in the same work, would you have the same suspicions. Because we do not have strong AI detection tools currently, it is important to acknowledge that a significant amount of intuitive evaluation is needed to identify red flags.
  • Use of detection tools. Cautiously deploy the use of AI detection tools with the complete understanding that these tools are not foolproof and can produce both false negatives and false positives. One or two checks through an AI detection tool should be a part of a broader strategy rather than on absolute proof. When using detection tools and your own intuition, be careful of falsely accusing students. Using AI detectors can provide you with some additional information, but it should never be used as absolute proof of a student’s AI use. Also consider the biases of the AI detector tools, such as their tendency to misidentify neurodivergent writing and ESL writing as AI writing. 
  • Ask a student to meet with you. When you request the meeting, use nonconfrontational language. You may want to say something like, “I’d like to meet with you about your assignment. I have some concerns.” When meeting with the student, it is a good idea to not be accusatory. It is important to remember that you can almost never be completely certain of their AI use. It is a good idea to ask them to share their writing process with you. You can ask them how they felt about the assignment and what the assignment initially brought to mind for them. You can ask them what was challenging about it and also what their process was. You can also ask about what resources they used. You can also ask them about the knowledge demonstrated in their writing. This can help you to determine whether the student has genuine knowledge and understanding of the content. You should then state your concerns clearly, such as “I am concerned about your writing for this assignment because it does not seem to match what I have received from you in the past. There were also a number of red flags in your writing that fit with AI writing. Additionally, the AI detector flagged your work for AI use.” Then you should point out the red flags and inconsistencies.  
  • Be aware that the student may not understand that their use was inappropriate. Many students use Grammarly Premium and other grammar-based tools and these tools can be used to rephrase students’ original writing. Students may not understand that this is not allowed. A powerful prevention technique is having several conversations with students at the start of the semester and then throughout the course about what is appropriate and inappropriate use with an emphasis on tools that students may not see as being AI.
  • Apply the policy in your syllabus. When developing these policies, you should consult with your department to get a feeling for what a reasonable penalty would be or perhaps to create a departmental policy together. Please do remember that we are in a early stage of this new technology and so it may be worth considering giving students the option to redo the assignment if they are honest and use it as a learning opportunity.  

 

Using ChatGPT and AI Generative Technology:

Resources: Below you will find a selection of resources to understand the potential and pitfalls of AI in academic settings. From generating dynamic content to fostering personalized learning experiences, AI tools offer a variety of opportunities for educational enhancement.

 

Ethical considerations

AI brings some new ethical questions to the table as we consider our own approach to this technology and our students’ use. We have a number of resources below that may help you in identifying some of these issues.

 

Calls to Action

  • Engage with Us: Participate in our workshops and webinars to explore the cutting-edge of AI in education.
  • Explore Further: Access our extensive collection of AI tools, educational resources, and ethical guidelines.
  • Share Your Experiences: Contribute to our collective understanding by sharing your insights and strategies for integrating AI into your teaching practices.

By fostering an informed and ethical approach to the use of AI technologies, we can enhance the educational experience at Moraine Valley Community College, ensuring that we remain at the forefront of academic excellence and innovation.