Applying AI to Education
Expressing Ideas More Effectively Through Writing
MyScribe (also known as Docuscope) was created by Suguru Ishizaki and David Kaufer in the Department of English to create an AI-powered intervention that can assist students as they begin to craft an essay. This system helps students reduce cognitive load to improve their ability to focus on expressing their ideas. Over the past two decades, MyScribe has been fine-tuned by scholars and students alike to improve the program’s theoretical framework. Today, the project encompasses a range of tools from computer-aided text analysis to technology-enhanced writing instruction. This program has been used by the Rand Corporation to analyze social media, by Education Testing Service to evaluate GRE essays under different scenarios and in scholarly studies.
Read More about MyScribe/docuscope
Personalizing Student Instruction and Feedback
Cognitive Tutors are software based on scientific learning that provide direction for students as they work through problems and provide context-sensitive, just-in-time help. Dietrich College faculty members John R. Anderson and Kenneth Koedinger are among CMU cognitive psychologists and computer scientists who pioneered this technology focused on "learning by doing." In 2003, the group launched Carnegie Learning, a CMU spinoff company that has brought Cognitive Tutors to thousands of schools. Powered by advanced technologies, including speech and handwriting recognition, facial analysis and machine learning, the program collects data from its interactions with the user to allow the cognitive tutors to adapt to individual students and allow the innovative data mining tool to evaluate and refine its design and functionality continuously. These tutoring systems then intervene to target students with multiple representations, scaffold their hint-seeking behaviors or targeting them with other forms of metacognitive support.
Learn more about Cognitive Tutors
Dietrich Analysis Research Education Ecosystem
Dietrich Computing and Operations, led by Vince Sha, associate dean for IT and operations, has initiated the development of a collection of tools that leverages existing and emerging large language models (LLMs) to solve specific use cases for faculty and staff members, including teaching, research and administration applications.
At this time, the collaborative initiative has several prototype projects in process. Once fully developed, the tools will be released as open-source projects, furthering the initiative's mission of democratizing access to advanced AI capabilities in academic settings. Tools in development include digital (Socratic) textbooks, Robocrates, Sway and a platform simulating democracy in the classroom.
Digital (socratic) textbooks
These AI-powered educational tools deliver personalized content and tutoring through conversational interactions. These textbooks adapt to each student's pace and interests, using the Socratic method to ensure comprehensive understanding. This new approach provides instructors with detailed insights on individual and class-wide learning progress, enabling targeted interventions and curriculum adjustments.
Collaboration with Daniel Oppenheimer, professor, Department of Social and Decision Sciences
Robocrates
These debate/argumentation bots are designed to deepen students' understanding of philosophical arguments and improve critical thinking. The bots engage students in Socratic dialogues, challenging assumptions and encouraging logical reasoning. They can simulate various philosophical perspectives to broaden students' understanding.
Collaboration with Simon Cullen, assistant professor, Department of Philosophy
Sway
This innovative AI-facilitated group chat platform uses advanced large language models to mediate and guide real-time discussions, aiming to reduce polarization and enhance mutual understanding. Sway employs two key AI components: personal discussion coaches and a discussion guide. These components work together to improve the quality of discourse. This groundbreaking platform represents a significant step towards using AI for educational and social good, with potential applications extending beyond academia to public forums and conflict resolution scenarios.
Collaboration with Simon Cullen, assistant professor, and Nicholas DiBella, post-doctoral research fellow, Department of Philosophy
Simulating diplomacy in the classroom
This innovative platform is designed to enhance strategic decision-making skills through immersive, realistic scenarios. Using cutting-edge technology, it leverages Language Learning Models (LLMs) and Retrieval-Augmented Generation (RAG) to create a multi-agent, multi-LLM environment for studying human-AI interaction and improving analytical reasoning. Players interact with a central white box system and collaborate with four LLM-based advisors: diplomatic, military, economic and intelligence. This simulation platform will be deployed as part of a pilot program on campuses across the United States and will serve as a testbed for exploring team-building dynamics, analytical reasoning, and the potential of AI-assisted decision-making in high-stakes environments.
Collaboration with Rafael López, deputy director for security policy studies, Carnegie Mellon Institute for Strategy and Technology