Skip to content

MABASA / LINKEDIN

Research Assistant - Biases in Humans and AI (Multiple Positions)

National University of Singapore

Location
Singapore, Singapore
Work type
fulltime
Source publication date
Original publication date not confirmed
Source closing date
Not supplied — check original advert

Job description from the source

Job Title: Research Assistant - Biases in Humans and AI (Multiple Positions) University-Level Unit: Arts & Social Sciences Faculty/Department-Level Unit: Economics Employee Category: Research Staff Location\_ONB: Kent Ridge Campus Posting Start Date: 03/11/2026 **Job Description** The Department of Economics, Faculty of Arts and Social Sciences (FASS), National University of Singapore, invites applications for multiple full-time Research Assistant positions on the research grant “debiaSING: Finding and Fixing Biases in the Human-AI Complex”, led by Prof Lorenz Goette. debiaSING brings together philosophy, behavioural science and computer science. It measures biases in large language models (LLMs), studies how these biases affect human users and how biased users affect AI, and develops interventions to debias both. The project will also build the debiaSING Machine, a public website where users experience AI bias first-hand. More details: https://www.ssrc.edu.sg/projects-awarded/lorenz2025/ We are recruiting for two profiles. Applicants should state which one they are applying for: * Profile A, Behavioural science: a degree in Economics (or a related field) and an interest in AI. * Profile B, Computer science: a degree in Computer Science (or a related field) and an interest in interdisciplinary research in behavioural science. **Scope And Responsibilities** **Profile A, Behavioural science** * Design and run online experiments measuring how LLM output affects human beliefs and decisions * Statistical and econometric analysis of human and LLM behaviour, including power analysis and preregistration * Analyse treatment effects of debiasing interventions, including heterogeneity across users * Program surveys and experiments; conduct literature reviews; contribute to papers and reports **Profile B, Computer science** * Build pipelines that generate test cases for biases (e.g. overconfidence, framing effects, gender bias) and run them across LLMs through their APIs * Run explainability analyses on LLMs to locate where biases arise in the model * Develop the debiaSING Machine website, its database of LLM responses, and an interactive visualisation tool * Work with behavioural scientists and philosophers to turn concepts of bias into tests; contribute to computer science conference papers **Both profiles:** Research Assistants work with at least two investigators across disciplines and take part in regular work-in-progress seminars. **Required Skills And Interests** * Profile A: strong grounding in statistics or econometrics; R and Python; interest in AI * Profile B: proficiency in Python; interest in interdisciplinary research with social scientists. Experience with LLMs (fine-tuning, training or interpretability) is an advantage. * Both: meticulousness, ability to work independently, good written and spoken English **Qualifications** * Profile A: Bachelor's or Master's degree in Economics (or another quantitative social science) * Profile B: Bachelor's or Master's degree in Computer Science (or a related field) **More Information** Full-time position, Kent Ridge Campus. One-year contract, renewable. Applicants should send the following by email to Prof Lorenz Goette (ecslfg@nus.edu.sg): * A cover note (no more than 500 words) stating which profile they are applying for * A two-page CV, with details of two named referees * A sample of work and/or code, if applicable Applications will be reviewed on a rolling basis until the positions are filled. Only shortlisted candidates will be contacted. Req ID: 34900

Check the original advert before applying

Apply through the employer or original publisher. An external site may require registration. Opening an advert does not submit an application.

Content is supplied by an external publisher and may change. For source and availability limits, read our listing disclaimer. How to check an original listing ↗