This project investigates whether students are already responding to AI by shifting away from fields like law, data analysis, and creative writing toward less automatable disciplines. You will use a novel measure of education-level LLM exposure and test it against UK university application data from 2020–2025.
About the project
Generative AI and Education Choices Large language models are already influencing labour markets, with evidence showing that highly exposed firms and occupations have reduced hiring since the introduction of ChatGPT in November 2022 (Brynjolfsson et al., 2025; Hosseini & Lichtinger, 2025; Klein Teeselink, 2025). This raises the important question whether young people are adapting to this new reality. This project aims to investigate that question by studying whether pupils, who are making their educational choices, are shifting away from fields like creative writing, legal studies, and data analysis toward less automatable disciplines. This is a fundamentally important question, as it gives us insights into the future of skills supply and human capital formation in the economy.
What You'll Do You will use a novel measure of education-level LLM exposure created by the mentor, which links educational credentials (degree subject, level, graduation year) to subsequent career paths, and calculating how "exposed" the career tracks of each educational program are to automation by LLMs. The core empirical work involves obtaining and analysing UK university application data from 2020–2025. You'll test whether applications to high-exposure fields have declined relative to low-exposure fields. There are potential extensions to the demand for microcredentials, vocational programs, and other types of educational choice.
Why This Matters Most research on AI and labour markets focuses on demand: which jobs are disappearing, which firms are hiring fewer workers. There is much less research looking at supply. Yet if students are already pivoting away from automatable fields, this has major implications for: • Future skill shortages and surpluses • The returns to different types of education • Retraining programs
What You'll Gain • Experience working with large-scale education and labour market data • Hands-on application of causal inference methods (difference-in-differences) • A potential publication-quality output on an important policy-relevant question • Mentorship from an active researcher in the economics of AI
Ideal Background MSc Economics or related field. Comfort with R preferred, Python also acceptable. Interest in labour economics, education economics, or the economics of technological change. No prior AI expertise required.
References Brynjolfsson, E., Chandar, B., & Chen, R. (2025). Canaries in the coal mine? Six facts about the recent employment effects of artificial intelligence (Working Paper). Stanford Digital Economy Lab. Hosseini Maasoum, S. M., & Lichtinger, G. (2025). Generative AI as seniority-biased technological change: Evidence from U.S. résumé and job posting data. SSRN. Klein Teeselink, B. (2025). Generative AI and labor market outcomes: Evidence from the United Kingdom. SSRN.
Theory of change
Understanding how AI reshapes labor markets is essential for anticipating and managing societal disruption from increasingly capable systems. Most research on AI and labour markets focuses on demand: which jobs are disappearing, which firms are hiring fewer workers. There is much less research looking at supply. This project will provide early empirical evidence on whether skill supply is already adapting to AI. This information can inform policymakers designing education systems, retraining programs, and deployment governance. If students are (not) systematically avoiding AI-exposed fields, this could signal either rapid behavioral adaptation or potential future skill mismatches.
Your role
Obtain educational choice data Merge educational choice data with education-level AI exposure data Run econometric analyses Write report
Prerequisites
MSc Economics or related field. Comfort with R preferred, Python also acceptable. Knowledge of econometrics and causal inference methods (difference-in-differences). Interest in labour economics, education economics, or the economics of technological change.
Location preference
I typically do meetings between 9AM and 5PM UK time, so as long as I can meet people during that window, it does not matter where they are located.
Application question(s)
- Provide a methodological critique of the following paper: https://papers.ssrn.com/sol3/papers.cfm?abstract\_id=5516798
About the mentor
Bouke Klein Teeselink is an Assistant Professor in Economics at King’s College London and a Senior Research Fellow at the AI Objectives Institute. His recent research focuses on the effects of generative AI on labor markets. In particular, his current work is looking at how generative AI is changing employment and wages around the world, how future advances in AI capabilities will shift occupational expertise requirements and employment outcomes, and how AI agents change consumer decision-making.