A chatbot could assist college students by providing instant feedback on assignments, offering personalized study tips for lectures, and even helping them stay on track with exercise routines. Scientists at Germany’s Technical University of Munich (TUM) have created a cutting-edge chatbot, dubbed Iris, designed to provide personalized support to computer science students struggling with complex programming tasks. A landmark study has conclusively validated the effectiveness of Iris, verifying its ability to enhance students’ comprehension of programming concepts and serving as a valuable adjunct to human instructors.
More than a thousand college students crammed into a packed auditorium. It’s not an uncommon sight in introductory informatics lectures. Since 2016, Stephan Krusche, a professor of Software Engineering, has led a team in developing the Artemis learning platform to cater to the unique needs of each individual student under these specific circumstances. While bearing similarities to established learning platforms, our innovative approach offers unparalleled opportunities. Moreover, college students don’t merely receive their assignments via the platform, but also gain instant feedback when they commit mistakes or demonstrate a limited grasp of concepts.
In October of the previous year, Stephan Krusch’s team at the Technical University of Munich (TUM) Centre for Educational Technologies successfully integrated a chatbot into their learning platform. The Iris app relies on the powerful GPT-3 massive language model developed by OpenAI. The platform is designed to support college students in addressing specific challenges, tailoring its resources to meet each individual’s unique needs and requirements. To seamlessly integrate with the training environment, the chatbot functions as an esteemed instructor. Rather than immediately proposing a definitive solution or specific actions, Iris provides subtle suggestions or poses key queries. This approach fosters independent critical thinking in college students. To boost study effectiveness, the AI exclusively addresses queries directly pertinent to the learning material.
To assess the effectiveness of Iris, researchers conducted a survey among more than 200 college students enrolled in three introductory informatics courses, notes Patrick Bassner, a doctoral candidate within the Department of Software Engineering. The goal aimed to investigate the perceptions of college students regarding the efficacy of Iris’s assistance, exploring whether they feel more at ease posing queries to a chatbot compared to a human tutor, as well as their subjective experience with Iris.
Scholars universally acknowledge Iris’s value as a valuable aid, yet many respondents still rely heavily on human tutors. The chatbot effectively comprehends queries and fosters a sense of comfort and security among college students when they utilize its services. With Iris, users experience a sense of profound intimacy, as if their conversations remain confidential and they’re spared the anxiety of being scrutinized for their queries. In large lecture settings, this feature shields college students from feeling compelled to overcome their hesitation and raise questions publicly before the class. As students increasingly rely on digital tools to prepare for exams, utilizing Iris has become a ubiquitous practice. Despite this, college students typically do not feel confident enough to tackle exam questions without consulting the chatbot’s guidance.
According to Stephan Krusche, the integration of Iris with Artemis demonstrates the value of AI-powered tutoring in higher education settings. “They offer tailored support that enhances learning outcomes and empowers college students to develop self-sufficiency skills.”
Researchers aim to further refine the chatbot Iris, seeking opportunities to deploy its capabilities across various subject domains.
