An AI-Enhanced Education

While Artificial Intelligence (AI) has been around for decades, the entry of ChatGPT in 2022 marked the beginning of a new era in AI. Institutions of higher education were particularly impacted by the explosion in AI use and are now navigating how to incorporate these technologies in a way that can improve educational models and help students to become ethical and responsible users. 

Columbia Engineering has worked to encourage faculty to think critically and broadly about how AI can be incorporated into various aspects of the student learning experience. This past spring, a call for teaching proposals was issued in the area of generative AI to inspire creative ideas from faculty. Winning proposals emphasized going beyond efficiency and automation to improve learning and assessment and to prepare students for a rapidly evolving professional landscape in the age of AI.

"Generative AI is no longer optional in computing education—it’s foundational," said Dean Shih-Fu Chang. "By embedding these tools directly into how we teach and how we structure coursework, we're preparing students not just to understand technology, but to shape it."

Image

Using GenAI to improve research, instruction, and student outcomes

In total, seven proposals were funded this cycle, representing a broad range of disciplines and approaches.

Lecturer in the Discipline of Chemical Engineering Chris V.H.-H. Chen received funding for his proposal, “An MEB Novice Chatbot to Enhance Chemical Engineering Critical Thinking Integrating GenAI into Engineering Education.” The project will extend use of the Material and Energy Balances (MEB) Novice Chatbot developed with the support of Vishal Misra and Karim Daouk and launched in Fall 2024. The team aims to develop AI models of deliberately reduced performance that simulate student difficulties in understanding concepts and reverse the role of students to become teachers teaching the error-prone AI models. This way, students learn more effectively by teaching AI models how to improve.

Tony Dear, senior lecturer in computer science (CS), has proposed a project on “Generative AI for Individual Student Feedback in Discrete Mathematics” that will use generative AI to better prepare students for the course content in Discrete Mathematics, a required course for CS students that depends on a great degree of mathematical preparation. The project aims to increase student engagement, learning, and performance through individualized student feedback via generative AI. 

A group of faculty led by Lecturer Lauren Heckelmen in biomedical engineering proposed “GenAI Integration in Required Junior-Level Undergraduate BME Courses.” By implementing generative AI into a set of courses for juniors, the team will be able to use the courses as comparison controls to quantify the impact of GenAI and apply their findings to future instruction and course development. An AI course assistant will provide students with assistance at any time. 

In the Department of Industrial Engineering and Operations Research, Assistant Professor Kaizheng Wang’s project “Digital Twins of Columbia Engineering Students” will create digital twins that identify systemic trends in order to better understand and emulate student needs. The project will produce testbeds for the research community that will enable the development of next-generation AI tutors without exposing real students to unproven systems. 

In chemical engineering, Assistant Professor Asher Williams’ project “Integrating Generative AI to Enhance Therapeutic Systems Design Education” is designed to prepare students for a future as researchers, teachers, and professionals and to support design-thinking and scientific communication. Students will incorporate GenAI to help them accelerate tasks such as literature mining, hypothesis generation, and regulatory writing, and will learn best practices related to the ethical and effective use of AI in research and writing. 

“Using Generative AI in Game Theory,” a proposal from Jay Sethuraman, professor of industrial engineering and operations research, will explore the use of generative AI tools in the course Game-Theoretic Models in Operations (IEOR 4407). The proposal will use GenAI to improve students’ understanding of the ideas and concepts that pervade game theory through an “agent” that students can interact with and learn from, and to explore cooperation between human players and AI agents. 

Computer Science Senior Lecturer Nakul Verma’s project “A Socratic AI Dialogue System for ML Theory Comprehension” will use generative AI technologies to facilitate guided inquiry and adaptive questioning rather than to provide answers so that students deeply engage with the subject matter. The project will create an innovative Socratic AI dialogue System designed specifically to enhance student comprehension of complex concepts in theory-heavy classes like Machine Learning (COMS 4771). 

While faculty pursue these projects across departments at the Engineering School, the Department of Computer Science has been evaluating the needs of students enrolled in the key CS programming course sequence. A group of faculty members in the department is revising Columbia’s introductory sequence for computer science majors to be more AI-focused. The courses under review included Introduction to Computer Science and Programming in Java (COMS W1004), Data Structures in Java (COMS W3134), as well as courses for non-majors such as Computing in Context (COMS W1002), Introduction to Computing for Engineers and Applied Scientists (ENGI E1006), and Intermediate Computing in Python (COMS W2132). 

The faculty plan to evaluate how different policies around AI usage affect student outcomes. The team will also explore how learning objectives evolve when AI becomes a built-in tool, and what strategies are most effective in preserving deep learning of programming fundamentals.

“What impressed us about these projects was not just the technical creativity, but the pedagogical thoughtfulness,” said Vishal Misra, vice dean for computing and AI of Columbia Engineering. “How do we teach students to be critical thinkers, responsible users, and fluent collaborators with AI? That’s the goal.”

Fall 2025 marks the launch of these GenAI-enhanced courses, with the goal of refining them further for the 2026–27 academic year based on feedback, assessment data, and continued student engagement.

AI as tutors and creative partners

Some engineering faculty have already been incorporating GenAI in the classroom prior to the most recent call for proposals. With support from Misra and his team, Professors Helen Lu, Adam Cannon, and Chris Chen have incorporated AI tutors into their pedagogy to provide ongoing, round-the-clock, customized course-specific support for students. 

Lu, senior vice dean of faculty affairs and advancement, has successfully used the AI tutor, set up with the help of teaching assistant Thomas Bina, with graduate students this past fall and plans to use it again this coming year with junior-level students. The AI tutor has been able to customize the learning and studying experience for students to aid comprehension. 

“The beauty of the AI tutor is that it adapts readily to the particular learner,” said Lu. “Students can calibrate the learning experience so that it matches their current level of comprehension. Students still attend office hours, but the discussion is often more in-depth and informative. In that respect, it’s enormously helpful for both teachers and students.”

Two courses putting AI at the center were also launched this past academic year and will continue to be offered. Last fall, Teaching Professor of Computer Science Adam Cannon and faculty from Engineering, philosophy, music, and literature launched AI in Context, building on Cannon’s popular Computing in Context course, which explores computing in other disciplines such as the humanities, finance, and economics. In a similar vein, AI in Context explores the social impact of AI and how students can be responsible users of AI. Bringing together faculty from different schools at Columbia, the five-module course also teaches students about the history and philosophy of AI. 

Lydia Chilton, associate professor in computer science, developed and teaches a course on AI and storytelling with Lance Weiler, associate professor of professional practice at the Columbia University School of the Arts and director of the Digital Storytelling Lab. The class, which has garnered much interest from students, invites them to use AI to share stories and experiences in new ways. Run like a studio class, students create and give feedback on each other’s work. They are joined by creative professionals who also wish to explore AI and storytelling. Chilton and Weiler plan to teach a follow-on course in the Fall, to be offered jointly by Engineering and the School of the Arts, where the students will use AI to build an immersive exhibit which will go on display in Miami after the class is over. Chilton is also co-teaching with Professor of International and Global History Matthew Connelly a class that integrates AI with Columbia’s Core Curriculum attached to "Contemporary Civilization." 

“In today’s landscape, innovation in how we educate students is just as important as innovation in other areas,” said Dean Chang. “As AI continues to transform education, Columbia will explore new approaches in curriculum and pedagogical programming to support students to become future responsible leaders.”

Lead photo credit: Jonh Abbott