Recent developments of large language models (LLMs) have revolutionized social science research. In particular the integration of LLMs in the process of data collection as well as the substation/amendment of survey data by silicon data (synthetic data) represent potential applications of AI in survey research. So far only few studies have systematically explored the consequences of using AI to collect data from human respondents. Also, few studies have suggested to generate silicon data and use them instead of data collected from human respondents. We invite students from around the world to join our group to study the pros and cons of such applications in a field experimental setting.