I am now using machine learning and natural language processing to answer sociological questions. My current work draws mainly on large text corpora (e.g., COCA, COHA, and Google Books Ngram), global news and bibliometric databases (e.g., GDELT and OpenAlex), and cross-national surveys (e.g., the European Social Survey and the European Working Conditions Survey).

My earlier work examined how flexible working, gig work and working-time arrangements shape well-being and gender inequality. I have studied the fragmentation of work time and its consequences for subjective time pressure and mental health, the ‘flexibility paradox’ in work schedules and locations, gig work and mental distress, and the gender division of labour within couples. This strand of work has appeared in journals including Work, Employment and Society, Sociology, The British Journal of Sociology, and Social Science & Medicine.

My current projects fall into two strands. The first strand uses representation learning to measure culture and social structure. I use word embeddings and transformer-based language models to study cultural change and gender norms in large historical corpora, and I use deep generative models (e.g., variational autoencoders) to learn latent dimensions of social stratification from survey data. The second strand treats generative AI as an object of sociological inquiry. I study how AI literacy and acceptance are socially stratified across Europe (funded by the Oxford John Fell Fund), why women are less likely than men to use generative AI at work, and how working in the AI era affects well-being through cognitive overload, multitasking, and cognitive anxiety.