Quantitative Methods and AI

In recent years, artificial intelligence (AI) has increasingly transformed the landscape of education. Yet students need to learn how to use it ethically and effectively as they are getting ready for the job market. Previous studies (Choustoulakis, 2024; Romero, 2025) show the benefit of integration of AI for students in quantitative courses such as personalized learning, enhanced engagement, problem solving and statistical writing. However, they didn’t focus on how integrating AI into these courses could make them ready for evolving job market. To this end, I redesigned a high-stakes assignment following the SAMR (Substitution, Augmentation, Modification, Redefinition) model, utilizing AI as an iterative mentor rather than a simple productivity tool. By shifting the technology’s role from basic substitution to redefinition, I integrated AI into the policy memo proposal process to serve as a Socratic collaborator. This setup pushes students to deeply examine their topics and refine their reasoning through a continuous feedback loop, preparing them to effectively engage with AI as a professional collaborator in the evolving job market. This framework is highly adaptable for statistics, econometrics, or other quantitative courses where students must learn to communicate complex data through well-reasoned policy arguments.

How Did I Do It?

Similar to other instructors, my syllabus included a section outlining the use of AI in a Quantitative Methods for Public Policy course that I am teaching to Master of Public Policy students. Rather than prohibiting AI, I encouraged its responsible and ethical use as a learning tool and highlighted the potential negative externality of overreliance on AI, particularly its impact on critical thinking. At the start of the semester, I provided examples of productive AI use, such as obtaining simplified explanations, additional practice questions, and coding assistance, and administered a survey to assess students’ confidence in statistics and their frequency of AI use. The results indicated that 46% of students rated their confidence level below 5 out of 10, and only 44% used generative AI on a weekly basis. In general, students cited concerns about academic integrity, mistrust of AI-generated responses, and environmental impact.

Based on these findings and my goal of supporting students’ growth, ability, and confidence in both AI use and quantitative reasoning, I integrated AI into the policy memo assignment. In the Quantitative Methods for Policy Analysis course I teach, this major assignment involves drafting a policy memo based on real-world data and presenting it to the class at the end of the semester. Students select a policy question of their choice, identify an appropriate dataset, and apply the analytical methods covered in class to answer that question. Some examples of the topic include homelessness and access to affordable housing, the effect of SNAP on test scores, Examining the Relationship Between Public Housing and Disaster Risk, and Relationship Between EU Funding to Member States and Public Support for the European Union. This project requires both data analysis and clear communication of findings in a policy context and accounts for 24% of the final grade. The activity was designed to help public policy students engage more deeply with their research topic and question by using generative AI as a knowledgeable mentor rather than an answer provider. It aligns with Universal Design for Learning (UDL) principles and sits at the Modification level of the SAMR framework, as AI meaningfully transforms how students develop and refine their research questions.

I divided the project into four steps:

• Initial research question and data set
• Proposal
• Reflection activity
• Final paper and presentation

Step 1: Initial Research Question and Data Set

Students begin by drafting a preliminary research question related to a public policy issue of their choice. While students were thinking about their questions of interest, I also ask them to find a data set to use to answer their questions. For brainstorming their questions, they did not use AI. This is aligned by the MIT study that if we brainstorm (human-component) first and then use AI, the product will be superior.

Step 2: Proposal

In this step, students use a prompt that I provided for them to have a conversation with a generative AI tool to draft their proposal. They were required to use ChatGPT, Gemini or Claude for this exercise. Here are few notes to consider:

• Students were required to use study mode/guided learning/learning style mode on the GenAI. This mode provides step by step guide for learning purposes, rather than providing an answer at once.
• The prompt gave the role of a supportive mentor to AI and asked AI to check the quality of writing based on professional job market standards.

The prompt consists of five subsections that students are required to include in their proposal and asked them clarifying questions regarding each part. Using the prompt, they were able to evaluate the clarity and testability of the research question.

Step 3: Reflection Activity

After students submit their proposals, they complete a reflection activity in which they respond to three questions related to the prompt they used:

1. What Worked
   (What parts of the prompt helped you organize your thoughts or move your proposal forward?)
2. What Didn’t Work
   (What parts of the prompt didn’t help you reflect or made the discussion less focused?)
3. New Prompt
   (Rewrite the prompt here.)
   

By responding to these questions, students revisit and reflect on the prompt they originally used in their proposal. This process encourages them to think critically about the structure and clarity of the prompt, identify areas for improvement, and revise it accordingly. In doing so, students engage more intentionally with the prompt and consider how to strengthen its effectiveness.

Step 4: Final Paper

In the final step, after completing their papers, students use an instructor-provided prompt as a checklist within a GenAI tool. This serves as a structured guide to help them verify that all required components of the assignment are included in their final submission.

Equipping students with the skills to use AI effectively, and to write strong prompts that enhance their research, is an essential investment for higher education.

What Did I Learn?

Equipping students with the skills to use AI effectively, and to write strong prompts that enhance their research, is an essential investment for higher education. In this teaching practice, I introduced students to a structured process in which they analyzed a well-crafted prompt, critiqued it, and revised it to better align with their research needs. Students accessed the prompt only after submitting their research topics and accompanying datasets. As part of the AI-use guidelines, they were encouraged to question and push back against AI-generated suggestions.

Because some students were unfamiliar with AI or did not follow the instructions closely, several encountered predictable challenges. A few fell into common “AI traps,” such as being steered away from their original research topic or being encouraged to use nonexistent datasets. Others, motivated by impatience, accepted AI’s suggestions uncritically, which led to misguided or unsatisfactory results. There was also one case of academic integrity violation in which a student used another generative AI tool to answer the questions posed by the assigned AI.

It is a valuable practice that prepares students for the job market, strengthens their understanding of AI, and enhances their ability to use AI responsibly and effectively.

In contrast, reflections from students who followed the instructions carefully indicate that the activity deepened their understanding of their research topics, helped them consider multiple perspectives, and prompted them to engage more thoughtfully with the relevant literature. They also appreciated how AI supported a more professional writing tone.

This activity can be incorporated into courses that involve research, such as statistics, applied econometrics, and other quantitative methods and research methods classes. Overall, it is a valuable practice that prepares students for the job market, strengthens their understanding of AI, and enhances their ability to use AI responsibly and effectively.

Author Profile

Shirin Sabetghadam, PhD is a Professorial Lecturer at the Department of Public Administration and Policy and a fellow at the Institute for Applied Artificial Intelligence at the Kogod School of Business. As a CFE SotL Fellow, she studied the use of technology and the costs and benefits of technological devices in the learning environment.

References and Further Reading

Choustoulakis, E. (2014). Integrating Artificial Intelligence (AI) Tools into Teaching Mathematical Economics in Tertiary Education. European Journal of Open Education and E-learning Studies, 9(1): 158-176.

Romero, L.S. (2025). AI-Assisted Statistical Writing in a Quantitative Research Methods Course, National Institute on Artificial Intelligence in Society.

Vaccaro, M., Almaatouq, A., and Malone, T. (2024). When combinations of humans and AI are useful: A systematic review and meta-analysis. Nature Human Behaviour, vol. 8: 2293-2303.