« Back to Results

Research on AI in Economic Education

Paper Session

Sunday, Jan. 3, 2027 8:00 AM - 10:00 AM (EST)

Marriott Marquis Washington DC
Hosted By: American Economic Association & Committee on Economic Education
  • Chair: Scott Simkins, North Carolina A&T State University

Evaluating the Impact of AI-Assisted Learning on Students’ Understanding of Economics and Business Concepts

Sining Wang
,
Case Western Reserve University
Scott Shane
,
Case Western Reserve University
Yufan Dong
,
Case Western Reserve University
Heyu Xiong
,
Case Western Reserve University
Jie Ning
,
Case Western Reserve University
Leonardo Madureia
,
Case Western Reserve University
Richard Boyatzis
,
Case Western Reserve University

Abstract

This paper reports causal evidence from a randomized field experiment embedded in introductory accounting and economics courses that tests how different modes of generative AI assistance affect learning and perceptions of learning. Students were randomized at the student–topic level within session blocks to one of four conditions: no AI, AI that solves problems for them, AI that clarifies mistakes, and interactive AI learning. Each experimental session included a practice exercise in which students had access to AI-assisted tools depending on their treatment status, followed by a quiz exercise in which all students reverted to a level playing field, and a post-activity survey recording their perceptions of learning. We find no statistically detectable increase in quiz performance across AI modes. Two AI modes - solution provision and interactive assistance - significantly increase practice performance. Solution-providing AI additionally reduces student perceptions of learning, whereas interactive AI does not meaningfully change perceptions. AI that clarifies mistakes produces no detectable effects on quiz performance, practice performance, or perceptions of learning. Overall, the findings show that how AI is used matters more than whether it is used, and that the effects of AI assistance are consistent with changes in the timing and structure of cognitive work rather than changes in student effort.

Do AI Chatbots Enhance or Replace Student Learning? Evidence from Structured Chatbot Integration in an Intermediate Economics Course

Andrew Jonelis
,
Syracuse University

Abstract

The rapid adoption of generative AI tools in higher education raises a fundamental question for economics educators: do AI chatbots complement student learning by reinforcing understanding, or do they substitute for it by encouraging cognitive offloading? This paper presents evidence from an experiment in which structured, topic-specific AI chatbots were integrated into an intermediate economics course (Economics of Emerging Markets) at Syracuse University. Beginning in Summer 2025, the instructor embedded pre-built, topic-specific AI chatbots into the course learning management system as a required component of course assignments. These instructor-designed chatbots allow students to engage with specific course topics through guided question-and-answer interactions, providing a structured alternative to unrestricted use of general-purpose AI tools. The study compares student performance across two periods: a control period of three semesters before chatbot implementation (Summer 2024, Fall 2024, Spring 2025) and a treatment period of three semesters with chatbot integration (Summer 2025, Fall 2025, Spring 2026). The same instructor taught all six semesters using comparable syllabi, assessments, and grading standards. The primary research question is whether students in chatbot-integrated semesters demonstrate improved retention of course material, as measured by quiz scores, exam scores, and overall course grades. A secondary analysis examines whether the intensity of chatbot engagement (measured through learning management system usage logs) predicts stronger or weaker learning outcomes. The empirical strategy controls for student academic background obtained from institutional records, including cumulative GPA, class standing, major, and prior economics coursework. This paper contributes to the emerging literature on AI in economic education by offering evidence on the complement-versus-substitute question in a setting where AI use is structured and instructor-directed rather than ad hoc.

Beyond the Ban: How AI Encouragement Shapes Student Attitudes

Molly Espey
,
Clemson University
Eamon Espey
,
Clemson University

Abstract

Artificial intelligence (AI) is transforming education. AI can provide personalized tutoring, generate study guides, and summarize complex information. Yet concerns abound, ranging from increased cheating to decline in critical thinking skills and cognitive atrophy, leading many instructors to ban AI in their classrooms and even to discourage its use outside the classroom. This study evaluates how instructor encouragement and guidance of AI use affect student engagement with AI in two economics courses. Students in both classes were encouraged to use AI as a study tool. One of these courses, however, included a multi-stage research project for which students were provided with more detailed guidance in using AI to develop their work. These students were also required to submit AI disclosure statements for each portion of the project. At the end of the semester, students in both classes were more positive about allowing AI use in academics, specifically with clear guidelines. They also felt the course helped them learn to use AI more productively and ethically. The students with the research project, however, were significantly more positive than the students in the other class in all aspects, from allowing AI use in academics to development of their ability to use it productively and ethically. These findings suggest that deliberate faculty integration of AI use in coursework, and providing clear guidelines, can help equip students with ethical and productive AI skills necessary for the modern workforce.

Teaching Principles of Writing Using Generative AI in an Undergraduate Economics Course

Oskar Harmon
,
University of Connecticut
Sijia Chen
,
University of Connecticut
Matthew Brown
,
University of Connecticut
Adam Patterson
,
University of Connecticut
Tara Grealis
,
Western New England University
Paul Tomolonis
,
University of Connecticut

Abstract

We examine how structured use of generative AI affects student writing and learning in an undergraduate economics course, with attention to AI as a complement to, not a substitute for, student thinking. We study a course in which generative AI is incorporated into selected writing assignments through guided prompts and revision exercises. In the course, writing assignments account for 33% of the grade. The study examines whether structured use of AI affects students’ confidence in writing economic arguments, their perceptions of the usefulness of writing in economics, and whether students experience AI as supporting or replacing their own thinking. We use a pre/post survey design to measure changes in students’ self-reported confidence in writing, perceptions of the relevance and engagement of writing assignments, AI-related skills, and career preparation expectations. In the writing assignments, students first produce their own draft, then use instructor-designed prompts to ask ChatGPT for feedback on topic development, clarity, and revision. They revise their work, document the drafting process, and write a reflection on how AI affected their thinking and writing. The study addresses a practical question for economics instructors: when does AI support student thinking, and when does it begin to substitute for it? AI is used as a tutor and revision aid, while assignment requirements preserve student authorship, reflection, and responsibility for the final argument. The study therefore speaks not only to whether students use AI, but to how course and assessment design shape its educational value. The project contributes evidence on AI tutors, cognitive offloading, and assessment design in economics education, and offers one model for using generative AI in course assignments without giving up student authorship and responsibility for the final argument.

Discussant(s)
Venoo Kakar
,
San Francisco State University
Ashley Orr
,
Ohio State University
Scott Wolla
,
Federal Reserve Bank of St. Louis
Jadiran Wooten
,
Virginia Tech University
JEL Classifications
  • A2 - Economic Education and Teaching of Economics
  • A1 - General Economics