I am still trying to figure out ways to use GenAI to enhance instruction, not just as a “faster google” like some students described it to me, but in a way to really support learning. So I have been playing with Claude as a personal instructor for a couple of days. And I like it!

I asked Claude to have conversations with me with the purpose of giving me opportunities to practise my Swedish. After a couple of days, I asked Claude to summarise what we have been doing:

“In our interaction, we have worked together to improve your Swedish skills through a dynamic and interactive process. You have formulated sentences and asked questions in Swedish, giving you the opportunity to practice expressing complex thoughts and ideas. In turn, I have provided immediate feedback, corrected grammar and spelling, and explained nuances in language usage. Our dialogue has covered various topics, from everyday situations to more specialised discussions about your work in academic development. This has allowed you to expand your vocabulary and practice using Swedish in different contexts. Through this process, you have received continuous practice and feedback, which is crucial for language learning. At the same time, you have been able to reflect on the potential for GenAI in teaching, which ties into your professional interest as an academic developer.”

And I agree, we are having interesting conversations, both when it comes to language learning but also for my professional life!

Regarding practising my Swedish with Claude, Claude picked up on Norwegian influences in the first message I wrote already, and pointed out where I was using Norwegian words or spelling in subsequent messages (which I have been trying to get the Swedish teacher to do, or to find a mentor for, both without success). We also discussed differences between Norway and Sweden beyond language, for example that here in Sweden, cheese slicers tend to have a straight cutting edge, whereas in Norway they tend to have a wavy one. Thanks for generalising that observation that I made when I moved here!

When it comes to academic development, we discussed, for example, how GenAI could be used in technical subjects, not just as a tool that gives “correct” responses or helps in that way, but in a way similar to how it is challenging and correcting me. Here, Claude suggested several interesting options and gave examples:

• Conceptual Dialogues: The AI can ask questions about fundamental concepts and ask the student to explain them. For example: “Can you explain Ohm’s Law to me? How are voltage, current, and resistance related?”
• Problem-Solving Conversations: The AI can present a problem and then guide the student through the solution process by asking leading questions. For example: “We have a circuit with a voltage of 12V and a resistance of 4Ω. What is the next step to calculate the current?”
• Debugging Exercises: The AI can present an incorrect solution or code and ask the student to identify and explain the errors. For example: “Here is a piece of code that should calculate the area of a circle. Can you find the error and explain how to correct it?”
• Application Discussions: The AI can ask the student to explain how a certain concept can be applied in real-world situations. For example: “How would you use knowledge of heat transfer to design an efficient cooler for a computer?”
• Interdisciplinary Connections: The AI can encourage the student to see connections between different technical subjects. For example: “How does the concept of energy conservation in physics relate to the efficiency of electric motors?”
• Experiment Design: The AI can ask the student to describe how they would design an experiment to test a hypothesis in their technical field.”

Claude suggested some prompt strategies for students to create good discussions:

• Define the Role: Ask the GenAI to assume the role of an interactive study assistant within the specific technical subject.
• Specify Interaction Type: Specify that the GenAI should use a Socratic method, asking questions and encouraging the student to think critically.
• Specify Knowledge Level: Ask the GenAI to adapt to the student’s knowledge level and gradually increase the difficulty.
• Encourage Explanations: Instruct the GenAI to regularly ask the student to explain concepts and reasoning.
• Include Error Correction: Ask the GenAI to identify and correct any misunderstandings or errors in a constructive manner.
• Encourage Application: Ask the GenAI to regularly inquire how the knowledge can be applied in practical situations.”

An example prompt that Claude provided:

“Act as an interactive study assistant in [specific technical subject]. Use the Socratic method to guide me through concepts and problem-solving. Adapt to my knowledge level and gradually increase the difficulty. Ask questions that encourage me to explain my reasoning and correct any misunderstandings in a constructive manner. Also, help me see how the knowledge can be applied in practical situations.”

I think working like this is super helpful and fun. And even without explicit instruction, the tone of the outputs was always encouraging and friendly, and made me want to continue the conversation, even though I had to put in effort (a lot at times). I can see that this use could be super useful for students, too!

One point I am thinking about a lot though is what does it mean for us as teachers when GenAI tools are perceived so positive, encouraging, and ultimately also trustworthy. And when it is also always available, day and night, never gets annoyed or is too busy to answer, and doesn’t potentially make people feel embarrassed for asking stupid questions or making stupid mistakes. Why would anyone still seek out interactions with teachers, how does it change our role, or maybe better, how should we reinterpret our role, to use GenAI to really enhance our instruction and student learning?