Anticipating User Needs: Insights from Design Fiction on Conversational Agents for Computational Thinking

Overview

• Computational thinking and programming are challenging to learn
• Conversational agents and generative AI (genAI) have potential to facilitate this learning process
• Current genAI chatbots focus on professional developers, not educational needs
• Involving educators is critical for creating useful and usable educational tools

Plain English Explanation

Learning computational thinking and computer programming can be very difficult for many people. Conversational agents and generative AI have the ability to help make this process easier by providing personalized guidance, interactive learning experiences, and even generating code for students.

However, the current genAI-based chatbots that exist today are mainly focused on helping professional software developers, not students learning these topics for the first time. It's important to get input from teachers and educators when designing educational tools like this, to make sure they are truly useful and easy for students to use.

The researchers in this paper worked with 9 instructors to get their ideas on what kind of abilities a conversational agent powered by genAI should have to best support students learning computational thinking and programming. The instructors envisioned an agent that can guide students step-by-step through exercises, and adjust its teaching approach based on each student's background knowledge, skills, weaknesses, and learning preferences.

Technical Explanation

The researchers conducted design fiction sessions with 9 instructors to elicit their ideas on the desired capabilities of a conversational agent supported by generative AI (genAI) for teaching computational thinking and programming.

Participants envisioned an agent that can guide students through exercises in a stepwise manner, tailoring its teaching approach based on factors like the student's educational background, current skills and deficits, and preferred learning style. This personalized, adaptive guidance is intended to facilitate the notoriously challenging process of mastering computational thinking and programming.

The insights gathered from these design sessions can inform future implementations of conversational AI tutors focused on teaching these critical skills, helping to ensure the tools are truly useful and usable for students.

Critical Analysis

The research provides a valuable starting point for designing educational conversational agents powered by genAI. By directly involving instructors in the conceptualization process, the authors have ensured the resulting vision aligns with the real needs and preferences of those who would be using such a tool.

However, this was an early-stage design study, and further research is needed to translate these ideas into working prototypes and evaluate their effectiveness in practice. Factors like student engagement, learning outcomes, and ease of use will need to be rigorously tested.

Additionally, the paper does not address potential downsides or limitations of relying on genAI for educational applications. Issues around bias, lack of transparency, and over-reliance on the technology should be carefully considered. Developing Generative AI Chatbots: A Conceptual Framework for Higher Education provides a helpful framework for analyzing these challenges.

Overall, this research represents an important step toward leveraging conversational AI and generative models to improve learning experiences. With continued development and testing, such tools could significantly enhance the teaching of computational thinking and programming.

Conclusion

This paper explores how conversational agents powered by generative AI could be designed to better support students in learning computational thinking and computer programming. By involving instructors in the design process, the researchers have identified key capabilities such an agent should possess, like providing personalized, step-by-step guidance tailored to each student's needs and learning preferences.

While further development and testing is required, this research lays valuable groundwork for creating educational conversational AI tools that can make these challenging subjects more accessible and engaging for learners. As AI models continue to advance, incorporating educator insights will be crucial for ensuring the technology is leveraged effectively to enhance teaching and learning outcomes.