Doctoral Researcher Position in the HUMAIN Project

Tiivistelmä työpaikkailmoituksesta: This is a doctoral researcher position on the HUMAIN project studying how large language models compare to human expertise, with funding until 31 August 2030 and a start date negotiable but no later than 1 November 2026. The role involves developing the COMIG framework and contributing to empirical data collection, embedding- and feature-based analyses, LLM adaptation experiments, and reporting as part of the RAISE research group. Candidates should have a background in computer science, data science, language technology, or related fields and strong skills in empirical study design, mixed-methods analysis, and computational text methods, with English required and Finnish/Swedish and LLM expertise considered merits. The position offers a starting salary of approximately 2800 euros/month and applicants must apply by 16 August 2026 via the University of Helsinki recruitment system with a CV, a motivation letter (max 2 pages) including research ideas, and contact details for two referees. Tämä tiivistelmä on luotu tekoälyn avulla. We are seeking a highly motivated doctoral researcher to join the What Makes Us Human? Identifying, Evaluating and Modelling Specialist AI Competence (HUMAIN) project, investigating how large language models (LLMs) approximate, extend, or diverge from human expertise across domains. The starting date is negotiable but should be no later than November 1, 2026. The position is fully funded until August 31, 2030 (46–48 months, depending on the starting date). About the project As LLMs become increasingly embedded in education, work, and everyday life, understanding their relationship to human competence is both scientifically and societally urgent. Existing benchmarks largely measure accuracy or linguistic fluency, but overlook the tacit, contextual, and ethical dimensions that define real-world expertise. HUMAIN addresses this gap by developing the online Competence Imitation Game (COMIG), a novel, experimentally validated framework that allows domain experts to interrogate and evaluate human- and LLM-generated responses through structured question-answer-evaluation conversations. Combining computational and interpretive methods, the project will generate reproducible, cross-domain datasets of paired human–AI responses, analysed through embedding-based similarity measures, feature extraction, LLM-as-evaluator approaches, and qualitative interpretation. Domain experts are involved from the outset, helping to develop practitioner-informed metrics for imitation quality, contextual alignment, and pragmatic fit. Empirically, HUMAIN will conduct seven COMIG studies spanning everyday life, education, and work, engaging over 900 participants internationally. The project will also evaluate adaptation techniques, including prompt calibration, retrieval-augmented generation, and fine-tuning, to examine how LLMs can better align with domain-specific reasoning. For more information on COMIG, please see the article ”Introducing the Competence Imitation Game: A Research and Teaching Tool for Context-Specific AI Literacy” ( https://ceur-ws.org/Vol-4138/paper3.pdf ). About the position You will support the further development of COMIG and contribute to empirical data collection, analysis, and reporting. You will be a part of the Research group on AI in Society and Education (RAISE) and work closely with Associate Professor Linda Mannila, PhD Otto Segersven and a postdoctoral researcher to be recruited. Who we are looking for We welcome applicants with a background in computer science, data science, language technology, or a related field. Applicants with, for example, a master's degree in social sciences combined with a strong background in data science, computational methods, or computer science are equally eligible. The position requires: • Strong background in research methodology, including the design and implementation of empirical studies with human participants. • Strong analysis skills, including both quantitative and qualitative methods. • Strong background in computational methods, especially for working with text data. • Experience with mixed-methods research (quantitative and qualitative). • Excellent English skills. • Knowledge of Finnish and/or Swedish, considered a merit, as some studies will preliminarily be conducted in Finnish contexts. • LLM expertise, including prompt calibration, retrieval-augmented generation, and fine-tuning, considered a merit. In addition, according to the system of the University of Helsinki, all doctoral researchers belong to a doctoral programme. The successful candidate should apply to Doctoral Programme in Science and obtain the study right within the trial period of six months of their employment contract. Please check the admission periods and eligibility criteria at https://www.helsinki.fi/en/admissions-and-education/apply-doctoral-programmes/how-apply-doctoral-programmes . We offer • A world class academic community t