March 29, 2023

ChatGPT: Ideas on How to Teach, Learn, and Evaluate

This article is a translation of a text first published in Eductive’s French edition.

Whether it is to teach, learn, or evaluate, ideas on the use of artificial intelligence, more specifically ChatGPT, are already travelling within the education network, on all levels. Which ones are more relevant to the college level?

At the Collège Montmorency, we organized 2 training sessions in February 2023 to demystify this chatbot and explore together the possibilities of this tool to:

  • undertake administrative tasks
  • plan a class or an activity 
  • learn
  • evaluate the learning process while avoiding the risks of plagiarism, cheating, and fraud

During these training sessions, we opted for a co-constructive approach taking advantage of our collective intelligence. During the guided experimentations, we provided digital spaces to facilitate collaborative notetaking (with Miro and Padlet).

In this article, we present an overview of the collected ideas and further ideas:

  • on the one hand, in order to make use of ChatGPT by both students and teachers
  • on the other hand, to sense the dangers related to ChatGPT and avoid them, or to focus on its conveniences

Relevance of ChatGPT responses

The use of artificial intelligence (AI) to undertake a task (whether it is professional or academic) is not de facto a lack of integrity. ChatGPT can prove to be a powerful tool that can help us accomplish different tasks. However, its use will be relevant if we have the expertise to evaluate its response.

Since ChatGPT does not use real-time search on the internet and relies on a database with information last updated in September 2021 (for now), its responses could be out of date. However, ChatGPT datasets expand as people use the software since it feeds on all the conversations.

ChatGPT combines data with stats and algorithms, from different sources in multiple languages, to write texts by predicting the logical sequence of words. Its responses may be inaccurate, wrong, or totally nonsensical. Through a few experimentations, you will notice that the robot can prove to be creative!

In fact, you will observe that some biases and stereotypes will appear in the suggested answers. They are from ChatGPT programming context. The robot was however trained to reject inappropriate requests and to provide answers free of any judgments or violence. It often concludes its answers with a good-willed paragraph reminding you to verify its answers with specialists.

Planning a class or an activity with ChatGPT

When looking for ideas or information, we usually type in a request on Google, consult databases, visit the library, read articles, share with our colleagues, etc. Now, we can also ask ChatGPT to make suggestions specifying our needs. It is one more resource to inspire!

When the time comes to plan a class or a pedagogical activity, we could use the tool to:

  • organize content (in order of priority, in charts, etc.)
  • simplify difficult concepts
  • summarize an article in a specific number of words
  • rephrase texts according to certain limitations, for example, with a specific number of words, style, or audience (young adults, experts, dyslexics, etc.)
  • obtain suggestions to teach a concept or a method:
    • “What are the learning difficulties…”
    • “How to explain…”
    • “In which order to present…”
    • “Which learning activities to…”
    • “Give me examples of…”
    • “Create a plan to…”
    • “Create a chart about…”
    • etc.
  • find ideas to evaluate a concept or a method:
    • “Propose 5 questions on…”
    • “Generate task instructions on…”
    • “Create a formative evaluation activity on…”
    • “Give me evaluation criteria to…” 
    • “Develop an evaluation grid to…”
    • “Offer feedback on…”
    • etc.
  • create question and exercise banks or lists of topics
  • create learning starters or writing tasks to complete
  • prepare a step-by-step procedure to solve a problem or accomplish a task
  • create a wide range of activities on the same topic, create different versions of an evaluation
  • generate charts and mind maps
  • compare aspects by presenting them in a chart
  • format scattered information into bibliographical references according to specific guidelines
  • program macros for Moodle
  • etc.

Learning and studying with ChatGPT

If used properly, ChatGPT can become a virtual tutor able to:

  • answer questions
  • provide explanations
  • help study

The students will definitely discover by themselves the different uses of this chatbot to support them in their learning process. Before they adopt more or less effective practices, take the time to show them how to optimize the different tool possibilities and how to avoid receiving inadequate answers. Above all, the students will need assistance to develop their critical thinking to have the capacity to determine the relevance of the answers provided by AI.

In order to have the students use ChatGPT as a study tool, we could ask them to:

  • search for information and validate it by comparing it to other sources
  • identify and analyze mistakes in the answers suggested by ChatGPT
  • identify the biases and stereotypes in the answers provided by the tool
  • generate an essay or oral presentation outline and provide comments
  • get a 1st draft of a text to be revised or completed
  • translate a text in another language to clarify their understanding of French (allophone students)
  • use the tool as a virtual teammate who contributes to activities in small groups
  • generate summaries or memory aids and validate them to review the content
  • generate problem-solving tasks or case studies, as a practice
  • generate questions to study, then validate the answers
  • have ChatGPT revise a text and explain the corrections made
  • generate a text in the style of a famous author [in French] and argue on how it matches the author or not
  • etc.

Evaluating while avoiding the risks of plagiarism, cheating, and fraud

With the advent of ChatGPT, issues related to plagiarism, cheating, and fraud are even more alarming. Since the tool provides answers to prompts in a well-structured text, how can we determine if the work submitted by the students is from their own reflection or was generated by AI?

ChatGPT can give many different answers to the same prompt, so there is no point in using plagiarism detection software such as Compilatio. The arrival of ChatGPT forces us to revisit our evaluation practices.

To avoid fraud and plagiarism, we could tend to rely on in-class evaluations, without a computer. But how is ChatGPT different from a parent helping their child with schoolwork? In addition, an evaluation strategy that relies only on exams does not align with inclusive practices such as:

In order to get the most out of the time spent in class with the students, a better strategy to take advantage of our subject expertise would be to plan our evaluations outside of class time. Here are some ideas for that purpose:

  • Educate students on intellectual integrity as soon as they enter college and remind them several times throughout their studies (see Comité Plagiat et intégrité intellectuelle, 2020 [PPT; in French]; Pagé et Jolicoeur, 2015).
  • Promote critical thinking on a regular basis and show how to demonstrate it in your subject area. For example, we can use these 3 strategies to teach critical thinking:
    • teaching for thinking
    • teaching of thinking
    • teaching about thinking

    (see Bérubé, 2013; Bizier, 2020; Boisvert, 1996, 1997, 2005).

  • Require the students to mention in their assignment if ChatGPT was used as a starting point. Ask them to describe the work that was done from the answer given.
  • Test our instructions in ChatGPT before presenting them in class, to anticipate plagiarism. To which extent is the proposed answer right? What could be added to the instructions to make the use of ChatGPT useless? Are there any patterns in ChatGPT’s responses?
  • Ask ourselves what is missing from the AI-generated responses to ensure the evaluation reflects the student’s learning.
  • Get to know our students better to use an overall judgment on their learning by tracking their progress in a learning portfolio or a reflexive journal.
  • Diversify learning traces during a semester to follow the process closely, and triangulate the traces to use a better evaluative judgement (see École branchée, 2022 [in French]).
  • Evaluate using the 3Ps to get different perspectives on the learning process:
    • product (production or achievement)
    • work process (method leading to the final product)
    • person (behaviours, metadiscourse on the work, explanations of reasoning, justifications of choices, analytical reflection, self-critics of production, etc.)

    (see Côté, 2017; Leroux, 2015 [in French]; Mastracci, 2012).

    In the case of teamwork, a collective evaluation is usually used for the production part. For an individual evaluation, the point mentioned on learning traces is the best way to get a personalized picture of the student.

    The process can be evaluated individually or collectively, depending on the context and the needs in order to use a better evaluative judgement.

  • In class, submit the evaluation questions to ChatGPT and show the students how it is possible to dig deeper. The questions could also become the starting point of a reflection on the work or intellectual integrity.
  • Assess an oral evaluation with individual interviews or in small groups (see Cormier et al., 2020).
  • Ask students to provide examples coming from:
    • their practicum
    • their personal experience
    • their course notes
    • the news
    • etc.

    Or else, ask them to base their research on resources ChatGPT does not have access to:

    • recent resources
    • PDFs you provided
    • videos
    • podcasts
  • Ask the students to create or comment on the content that ChatGPT is unable to generate or analyze for the moment:
    • mind maps
    • diagrams
    • graphics
    • timelines
    • procedure videos
    • etc.

The unavoidable adaptation of our practices

If, in certain fields, AI tools increase and facilitate some processes, in education, the phenomenon is considered with some restraints. Chatbots such as ChatGPT are a reality that we can still choose to ignore, but that will catch up with us sooner or later. We better adapt now!

The teaching profession does not seem in danger: the pedagogical relationship remains one of the most significant factors of success (it influences motivation, involvement, well-being, etc.).

Instead of forbidding the use of ChatGPT, let’s do the opposite and let’s educate them on how to use the tool responsibly. Let’s help them develop their judgement and critical thinking from IA-generated content.

And let’s talk about it! Within our department, our service, our college, our network…What we know today will already be obsolete tomorrow. It is through the sharing of our experiences that we can follow the rapid development of artificial intelligence, in order to better understand educational issues and identify opportunities to support learning.

Technical tips for use

The participants in the 2 training sessions at the Collège Montmorency were asked to identify technical problems they had faced while experimenting with the tool, and to propose tips to use ChatGPT wisely. Here are some of them:

  • Contextualize the questions to get more precise answers.
  • Add details to develop an answer that is considered poor. Sub-questions are sometimes necessary to reach a satisfactory answer.
  • Set a limit to the number of words to get a shorter answer.
  • Use quotation marks to constrain the prompt to an expression. Otherwise, the robot considers every word separately.
  • Ask ChatGPT to provide the references of the information, and check and triangulate the results. The references provided are sometimes fabricated!
  • Click on “stop generating” to interrupt an answer generation. Otherwise, ChatGPT crashes. Refresh the page if it stops answering.
  • Archive your conversations and visit your history to retrieve responses.

Thanks to Philippe Lavigueur, technopedagogical advisor and colleague, for his support at the training sessions organized at the Collège Montmorency in February 2023, and for his contribution to this article.

Resources on the solutions proposed

Bérubé, M. (2013). Comment aiguiser le jugement critique de nos étudiants? Pédagogie collégiale, 26(2).

Bizier, N. (2020). Les questions scientifiques socialement vives : des controverses pour donner du sens aux apprentissages et pour réfléchir sur les rapports aux savoirs. Pédagogie collégiale, 33(2).

Boisvert, J. (1996). Développer la pensée critique au collégial. Pédagogie collégiale, 10(2).

Boisvert, J. (1997). Une stratégie d’enseignement de la pensée critique au collégial. Pédagogie collégiale, 11(2).

Boisvert, J. (2005). La formation de la pensée critique au collégial en sciences de la nature, design d’intérieur et soins infirmiers. Pédagogie collégiale, 18(3).

Comité Plagiat et intégrité intellectuelle (2020). Guide sur le plagiat et l’intégrité intellectuelle. Réseau REPTIC et REBICQ. [PPT; in French]

Cormier, C., Turcotte, V., Arseneault-Hubert, F. et Voisard, B. (2020, April 2023). The One-on-One Interview as an Innovative Assessment. Éductive.

Côté, F. (2017). L’évaluation des apprentissages au collégial: un réseau de concepts pour guider les pratiques évaluatives. Pédagogie collégiale, 30(4).

École branchée. (2022). L’élève se dévoile à travers les traces d’apprentissage. L’école branchée, 25(2). [in French]

Leroux, J. L. (dir.) (2015). Évaluer les compétences au collégial et à l’université: Un guide pratique. Association québécoise de pédagogie collégiale. [in French]

Mastracci, A. (2012). Présentation des outils pour l’évaluation des apprentissages en créativité. Cégep Marie-Victorin.

Pagé, M., et Jolicoeur, N. (2015). Prévenir le plagiat et la tricherie: pour une stratégie concertée et à long terme. Pédagogie collégiale, 28(3).

Suggested resource on AI and ChatGPT

Dubreuil, D. (2023, 5 février). ChatGPT et IA génératives. Pearltrees. [in French]

Duguay, S. (s.d.). IA en éducation. Padlet. [in French]

FADIO. (2023, February 22). #SemaineFAD2023 : ChatGPT, IA et formation à distance : une nouvelle ère de l’apprentissage. . YouTube. (60 minutes). [in French]

Gauthier, L.-M. (2023, February 6). ChatGPT: Artificial intelligence on the dorrstep of college institutions. Éductive.

Herft, A. (s.d.). A Teacher’s Prompt Guide to ChatGPT aligned with «What Works Best». @herfteducator. [PDF]

Lapolice, M.-E. (2023, mars). Apprivoiser ChatGPT: opportunités et défis en éducation. Webinaire APOP.

Martineau, J. (2023, 10 février). ChatGPT pour enseigner la poésie et la création littéraire. Éductive.

About the authors

Stéphanie Carle

Stephanie Carle is a professional doctoral student in education at the Université de Sherbrooke and she holds a Master’s Degree in Communication Studies. She has been working within the college network since 2001.

Lecturer for Performa since 2003, she is now a pedagogical advisor at the Collège Montmorency, after having worked for 6 years at the Cégep de Lanaudière à Joliette as an Automation teacher, both at the general and continuing education. From 2014 to 2022, she was also the chief editor for Pédagogie collégiale.

Stéphanie Carle is particularly interested in professional development, the relationship to knowledge, and educational approaches.

Xavier Martel-Lachance

Holding a Master’s Degree in Communication — specialized in video games and gamification at the UQAM, Xavier Martel-Lachance is interested in the practical aspects and interactivity of multimedia as well as communication theories and scientific research. His master’s thesis was on the study of group cohesion.

Along with his academic path, he was involved in a wide range of digital projects: game design, interactive installations, video mapping, and all kinds of artist coaching. He also maintains a programming-oriented artistic practice on real-time interactive multimedia content reacting with sounds.

Former lecturer at the music school Vincent d’Indy, he now works at the Collège Montmorency since fall 2021.

Notify of

0 Commentaires
Inline Feedbacks
View all comments