Instructor Profile: Nazanin Khazra

Nazanin Khazra is an applied microeconomist working on urban economics and applied machine learning. As a course instructor, Professor Khazra has integrated generative AI tools into assessments and activities, supporting an inclusive and engaging learning environment for students.  

Q: How have you integrated generative AI tools into course assessments, and how have these tools supported students’ AI literacy skill development and their understanding of course content?

Big Data Tools for Economists (ECO225) is a research-based course for second-year students. In this course, students learn cutting-edge tools; they learn to code and work with data; the outcome is individual research papers using real-world datasets. Developing research questions and understanding the literature they are working with are complicated steps for any researcher, especially undergraduate students. I used gen AI tools to design guided assignments to help with these two steps. It is important to mention that students were using ChatGPT for other parts of their work, such as coding, interpreting the results, and getting ideas, in addition to the two official assignments I discussed. Using Gen AI was invaluable in all parts of this course.  

In the research question development phase, students used ChatGPT to generate and refine potential research questions based on their datasets. This helped them explore a wide range of ideas quickly, giving them insight into what makes a well-constructed, feasible research question. Throughout this process, students were encouraged to critically assess and refine the AI-generated questions, contributing to deeper engagement with the course content and improving their ability to narrow their research questions to something feasible and exciting.   

In the literature review exercise, students learned to conduct thorough literature reviews by searching for key papers related to their research topics. They identified core themes and methodologies through databases like Google Scholar. Using Research Rabbit, they created visual cluster maps to understand how their research connects to broader academic discussions, facilitating an understanding of interdisciplinary relationships. Finally, they used ChatGPT to summarize essential aspects of each paper and organized this information into a table, serving as a foundational tool for their literature review.   

Q: In assessments that involve generative AI tools, how do you design tasks and articulate criteria/expectations, so that they specifically evaluate students’ unique human abilities (as opposed to generative AI capabilities)?

To ensure we are using Gen AI to complement and improve learning and not as a replacement, I emphasize tasks requiring critical thinking, creativity, and deep engagement with course content—skills that generative AI tools like ChatGPT cannot fully replicate. For example, in the literature review and research question development assignments, students use AI as a tool to support the research process, but their success in these tasks depends on their ability to critically evaluate AI-generated outputs, refine complex research questions, process their data, and interpret findings in a meaningful way.  

If students make any claims, they must back them up with data and present their findings using maps, plots, regressions, or machine learning outputs. While ChatGPT can help them understand the literature, develop new hypotheses, and refine their research questions, it’s ultimately their responsibility to provide evidence for their claims. This requires critical thinking, data analysis, and applying the methods discussed in class—all of which are challenging tasks. 

Using generative AI in my course is like having a flashlight while hiking in the dark—it doesn’t do the work for you, but it helps you see the way forward! This approach ensures that students’ unique intellectual contributions are the core focus of the assessment rather than the capabilities of generative AI tools. 

In the research question exercise, while ChatGPT assists students in generating initial ideas, the students themselves are responsible for refining these questions into more sophisticated research inquiries. For example, one student began with a basic question about street-level crime patterns but, through iterations with ChatGPT, refined their research focus to examine the impact of street lighting on nighttime crime. This task required the student to think beyond simple correlations, narrate a story around the data, and consider subgroup heterogeneities—skills that demonstrate their unique human ability to approach complex research questions creatively and critically.  

In the literature review assignment, students used AI tools like Research Rabbit to visualize connections between studies. However, the true value of the task lies in their ability to contextualize and interpret these visualizations. They had to critically assess how their research fits into the broader academic landscape, which demanded deep understanding, judgment, and knowledge of the literature—skills that cannot be automated. 

Q: Outside of assessments, how have you integrated the use of generative AI into course activities, interactions, and office hours? 

In ECO225, we use ChatGPT in other course activities and in collaboration hours. Many students use ChatGPT as a personal coding tutor, allowing them to troubleshoot issues and learn coding concepts at their own pace. This access to AI support has filled the gap, providing immediate assistance that was not feasible with traditional office hours. I previously aimed to improve accessibility for students needing extra support, but the limited TA hours made it difficult to achieve the desired results. With the introduction of ChatGPT, students now have constant access to assistance. ChatGPT can effectively fulfill the role of a TA in helping with coding problems to receive instant feedback. 

We use Gen AI in collaboration hours as well. Collaboration hours are structured sessions designed to promote teamwork and problem-solving among students, rather than traditional office hours which are focused solely on individual queries. I encourage students to attend these group sessions and bring coding or research questions and work together to find solutions. This approach creates a sense of community and motivates students to engage actively with the material. As a side note, these sessions are mandatory and are supervised by a TA. During these hours students frequently use ChatGPT to assist one another. As they work together to find solutions, they improve their understanding of course material while also building a supportive community.  

Q: Looking ahead, what goals do you have in terms of integrating generative AI into upcoming courses?

One task on my agenda is to teach students how to use these tools effectively and critically evaluate their outputs. Learning the practical and ethical use of these tools will prepare them to use AI responsibly in their academic work and future careers. For example, students could be assigned to generate data insights using AI tools and then analyze the results for potential biases. This exercise helps them understand the limitations of AI, as these systems can reflect biases from the data they were trained on. Understanding the limitations of AI models is important for promoting ethical research practices and teaches students that AI must be used carefully; what AI generates is not the truth; it’s merely a prediction which is subject to biases and error.  

Designing such activities is challenging and requires significant thought and effort. My goal is to develop practices that can be adopted by educators worldwide in the future.