Classroom Learning and Assessment Suite (CLAS)
The Classroom Learning and Assessment Suite (CLAS) is a set of artificial intelligence (AI)-driven solutions that complement in-class learning with interactive study and assessment. Primarily ideated by faculty interested in leveraging AI platforms to enhance student learning, the suite creates new mechanisms for both students and faculty members in supporting the creation of interactive learning assessments and streamlining grading of low-stakes student submissions, such as formative assessments, using generative AI-based platforms. CLAS is suitable for users of diverse programming backgrounds, from no programming background to seasoned programmers.
Motivating this project is a desire to augment and enhance student learning by enabling a 24/7 generative AI tutoring assistant based on the coursework and learning objectives provided by the instructor. This additional support for self-study fosters curiosity and independent learning, and feedback to the instructors regarding a classroom of self-study activities helps inform topical areas to emphasize during lecture.
The platform includes tools for both students and faculty by leveraging the strengths of generative AI platforms, including a seamless way for students to engage in instructor-guided self-study, tools for supporting instructors in automated creation of course materials and evaluation of self-study work submissions, and explanatory template code for instructors interested in providing hands-on instruction about generative AI in the classroom.
This innovative approach allows students to interact with any text or authoritative source, encouraging active questioning, answering, and exploration. The platform guides instructors in creating a repository of media including course texts, websites, and YouTube videos, which students are able to link to through an online interface. Users of the interface are additionally able to upload their own documents, get help with prompts, and view the source material specifically used by the generative AI to compose its questions and answers. Although the generative AI still uses information from its training, providing a browsable set of sources contributing to its response helps to combat misleading information due to hallucinations.
CLAS provides several mechanisms for using the platform – including a repository of thoughtful and tested prompts to be used with generative AI interfaces including ChatGPT, GPT4, Bard, and Claude; point-and-click user interfaces for expanded capabilities for creating new prompts and downloading conversational text; and interactive Python notebooks with single-click opening and interaction. The open source platform also provides a Python library which allows users to embed the functionality of the platform in their own academic or research work.
An augmentation already extending the self-study work is an oral exam component involves students answering questions verbally, enabled by text-to-speech capabilities of some AI models. The generative AI then provides feedback and preliminary scores, helping assess verbal expression of knowledge and thinking skills.
This project is part of the Data Science for Social Good (DSSG) program. Senior Data Scientist, Dr. Charreau Bell, is the technical team lead of the project and supports the following integrated, interdisciplinary team of code developers:
- Katrina Rbeiz, Ph.D. Student, Psychology, Vanderbilt University
- Minwoo Sohn, Graduate Student, Data Science, Vanderbilt University
- Ricky Sun, Graduate Student, Data Science, Vanderbilt University
- Eleanor Beers, Graduate Student, Data Science, Vanderbilt University
- Kevin Chen, Undergraduate, Computer Science, Vanderbilt University
- Adam Levav, Undergraduate, University of Maryland
- Varun Koduvayur, Undergraduate
The project’s PIs are:
- Jesse Spencer-Smith, Ph.D., Chief Data Scientist, Data Science Institute, Professor of the Practice, Computer Science, Vanderbilt University
- Jesse Blocher, Ph.D., Director of Graduate Studies, Data Science Institute, Associate Professor of the Practice, Owen School of Management, Vanderbilt University
- Yaa Kumah-Crystal, M.D., M.P.H., Pediatric Endocrinologist and Professor, Vanderbilt University Medical Center
- Charreau Bell, Ph.D., Senior Data Scientist, Data Science Institute, Assistant Professor of the Practice, Computer Science, Vanderbilt University (Staff Lead)