School of Business and Management Department of Management 216 Judgment and Decision Making in Policy Supervisor: David HAGMANN / MGMT Student: KO Sung Kit / COGBM Course: UROP 1000, Summer This progress report documents the initial struggles, design, development and implementation of a full-stack, cloud-hosted platform for conducting experiments regarding persuasion at the extensive margin. We attempt to investigate how psychological and social factors may influence an individual’s choice to engage in conversations that might affect their internal beliefs. Initial approach involved utilising the open-source platform deliberation-empirica, but was subsequently abandoned due to issues regarding minimal documentation, deployment complexity, and insufficient scalability. Research continued with the development of a platform substantially accelerated with the use of AI coding assistants, which lowers the barrier of entry for researchers with limited engineering backgrounds. Judgment and Decision Making in Policy Supervisor: David HAGMANN / MGMT Student: WONG Chung Wung Carlo / QFIN Course: UROP 3100, Fall This progress report presents an ongoing experiment investigating the impact of generative artificial intelligence (AI) assisting human interviewers on undergraduate admissions interviews. The latest iteration of the research question is inspired from an article about reducing misinformation through dialogue with artificial intelligence. (Costello et al). A proof-of-concept test will be conducted on randomised respondents simulating a job interview through dialogue with generative AI. We will introduce an alternate interview format alongside existing formats used by the admissions department, and interviewers will have the flexibility to choose the preferred format for each interview. While data collection is pending, it will be collected via feedback surveys and written notes from interviewers. If the feedback is positive, the pilot run will be conducted in the incoming JUPAS applicants. The findings will provide valuable insights for optimising undergraduate admissions processes despite the experimental constraints due to the potential impact of evaluating candidates. Judgment and Decision Making in Policy Supervisor: David HAGMANN / MGMT Student: YIN Wenhui / ECOF Course: UROP 1100, Fall The primary objective of this study is to investigate how the fundamental attribution error (FAE) manifests in the modern context of GenAI usage. Through a three-phase experimental design, participants engage in a debate-style writing task where some writers have access to ChatGPT while others do not. Independent audiences rate the arguments’ strength, and writers evaluate their opponent’s knowledge on the topic before and after learning about potential assistance. The study hypothesizes that knowledge of the opponent’s GenAI usage will lead to decreased perceptions of the opponents’ capabilities, regardless of actual writing quality. Results will contribute to our understanding of modern-day decision-making and could have implications for educational institutes where use of AI tools are increasingly prevalent.
RkJQdWJsaXNoZXIy NDk5Njg=