The Role of ChatGPT-4 in Shaping Dialogic Learning Experiences: A Practical Example

In my last article, I explored dialogic experiences where students can interact with generative AI tools through prompt engineering with the aim of, for example, facilitating metacognition and enhancing creativity and critical thinking.

Within the last week, UNESCO published its ChatGPT and artificial intelligence in higher education: quick start guide. The guide explores the different roles AI can have within a higher education context, which is related to designing dialogic experiences for students (see table below).

Having a role for AI implies that there is specific purpose or function that the AI is intended to fulfill. In order to fulfill these roles effectively, one could argue that it is necessary to have a method or approach for using AI in a way that achieves the desired outcome.?The method provides a structured way of applying AI to achieve the intended role.?

Using Mike Sharples ' experimentation with these roles for ChatGPT-4 as a starting point, I thought it would be good to provide a concrete example for what I was talking about in the previous article. Below is a description of my journey.

A dialogic experience with ChatGPT-4

I first adapted the prompt Mike Sharples used for his post. My aim was to define the chatbot's role using one of the suggestions in the UNESCO's guide.

I then answered ChatGPT's three questions. For the purpose of this experiment, I lied a bit: I have very little to no knowledge of Signal Processing and Communications; but let's assume I am an Electric and Computer Engineering student. Here's our conversation:

As a "student", I am often confused by templates and structures. Do I always have to follow a template? Some teachers in secondary school graded me poorly if I did not follow the essay templates they gave me for assignments; however, my current university language teacher said templates should be used only as a reference. Somewhat confused, I chose to ask ChatGPT4:

By this point, I (the student) have received new information and I felt it was time to test my understanding.

I didn't do so well on the quiz.

What results from this experience tell me

So what did I learn from this experiment? I suppose this dialogic experience could be used as a self-regulated learning activity to assist with a larger assessed task within a course, such as design a corporate technical blog. However, it may not be that simple. This is only a hunch, but I suspect students would need guidance on how to go through the process. It would be helpful if a teacher provides example prompts and guidelines for students if they wish to do self-regulated learning. This activity could also be embedded into lessons, and students pair up to go through the same process.

As I mentioned above: Having a role for AI implies that there is specific purpose or function that the AI is intended to fulfill. In order to fulfill these roles effectively, one could argue that it is necessary to have a method or approach for using AI in a way that achieves the desired outcome.?The method provides a structured way of applying AI to achieve the intended role.?I discussed some methods in another article but reference them here for quick access:

• Method 1: Using AI chatbots to generate analogies and examples that help students transfer their knowledge to new contexts and domains. For example, an AI chatbot can generate an analogy between the human circulatory system and a city’s water supply system to help students learn about biology.
• Method 2: Using AI chatbots to generate explanations that help students break the illusion of explanatory depth and deepen their understanding of concepts. For example, an AI chatbot can generate an explanation of how a microwave oven works to help students learn about physics.
• Method 3: Using AI chatbots to generate arguments that help students critically evaluate explanations and evidence. For example, an AI chatbot can generate an argument for or against a claim such as “social media is good for democracy” to help students learn about politics.
• Method 4: Using AI chatbots to generate questions that help students test their knowledge and identify gaps or misconceptions. For example, an AI chatbot can generate a question such as “What is the difference between a hypothesis and a theory?” to help students learn about science.
• Method 5: Using AI chatbots to generate creative content that help students apply their knowledge and skills in novel and engaging ways. For example, an AI chatbot can generate a poem, a story, a song, or a drawing based on a topic or theme that students are learning about.
• Method 6: Using AI chatbots to generate solutions that help students solve problem-based situations. For example, an AI chatbot can generate a solution for how to reduce food waste in a school cafeteria to help students learn about sustainability.
• Method 7: Using AI chatbots to generate prototypes that help students design design-thinking based situations. For example, an AI chatbot can generate a prototype for a wearable device that helps elderly people monitor their health to help students learn about engineering.

By assigning roles and methods to generative AI tools, such as ChatGPT-4, educators create a great potential for facilitating metacognition and enhancing creativity and critical thinking. Below are some of my thoughts on the different types of dialogic relationships that may have occurred in this experiment. For example:

1. Dialogue with ourselves (metacognition) through prompt engineering: When students craft their queries, they consider their goals, clarify their thoughts, and articulated their requests for ChatGPT-4 to take on specific roles. This self-dialogue facilitates metacognition and introspection, as they think about how to effectively frame their prompts for the desired outcomes.
2. Dialogue with selected human knowledge: As a generative AI tool, ChatGPT-4 is trained on vast amounts of data, encompassing a wide range of human knowledge. When it provided guidance on technical writing and offered quizzes to test the student's understanding, the student engaged in a dialogue with the collective wisdom encoded within ChatGPT-4. This interaction exposed the student to different perspectives and ideas, enriching their understanding of writing corporate technical blogs. Of course, we also need to consider implications related to bias and limitations of the chatbot; that's where the teacher becomes important.
3. Dialogue with social systems: By asking ChatGPT-4 to play the roles of a tutor, co-designer, motivator, and quizmaster, the student effectively embedded their (i.e. student and chatbot) interaction within a social system. In this context, the chatbot acted as a facilitator and enhancer of the student's learning experience, simulating different aspects of educational settings to provide a more comprehensive and engaging learning process. This was limited in the example I used in my experiment; however, I'm sure we could explore this idea more. For example, teachers could provide prompts or guidelines instructing students to simulate different social systems, such as within a company that wants to launch a blog to promote its products.
4. Dialogue with predictive texts (possibilities): As a generative AI tool, ChatGPT-4 provided context-sensitive information and suggestions in response to the student's prompts. By engaging in a dialogue with the predictive texts the chatbot generated, the student explored alternative ideas, suggestions, and approaches to learning and understanding technical writing for a corporate technology blog. This dialogic experience allowed the student to think beyond your initial assumptions and preconceived notions, broadening your creative thinking. In this case, it may be challenging their prior knowledge of composition and templates. Again, this was limited in the example I used in my experiment; however, I'm sure we could explore this idea more.

Final thought

Of course, it's important to address the potential limitations and ethical considerations of using AI tools like ChatGPT-4, such as bias in the training data or the reliability of the information provided by the AI. Knowing this, I would still stress the need for human supervision and guidance alongside AI-dialogic learning experience.