AI
& Behaviour

Two Research Assistants – Science in layman language

Department: Computer Science (AI & Behaviour group)
Work hours: 0.4 FTE for 11 months (Nov 2025 – Sep 2026)

About the project

For years, the library sector has been considering and examining the accessibility of scientific publications. With open access, more and more scientific articles are becoming freely available to everyone. A 2020 study (https://zenodo.org/records/4143313) shows that 40% of readers of scientific publications on SpringerLink are non-academics. Help with reading scientific publications is a frequently heard need. 

AI systems are widely used, but we have little insight into their reliability. In the search for understanding and access, existing AI solutions such as Chat-GPT, Elicit and Evidence Hunt are being used on an increasingly large scale. Knowledge about the reliability of these tools is still limited. A study by Utrecht University shows that LLMs are not able to match the level of manually written scientific abstracts (https://doi.org/10.1098/rsos.241776). There is still little knowledge about the effect of different ways of using LLMs (configurations and prompts) on reliability and comprehensibility for non-scientific users.

Therefore the Dutch Royal Library (KB), the VU Library, and the AI & Behaviour group want to investigate how to make open access scientific publications understandable and reliably accessible using Generative AI. We will perform a small-scale experimental study on the development and evaluation of open and reliable LLM-based configurations that can be used to translate scientific articles into understandable Dutch for different types of end users, such as healthcare professionals and policymakers. The development and evaluation will be done in close collaboration with the professional end users. 

For this project, we are looking for two master students that can work as Research Assistants. 

Key responsibilities RA 1 (background in AI or CS)

  • setting up and working with a technically functioning LLM platform (e.g. based on Nebula (https://networkinstitute.org/nebula/))
  • design and setup of LLM configuration
  • prompt engineering and processing input
  • collection of respondent output/data
  • reporting and contributing to scientific publication

Key responsibilities RA 2 (background in Social Sciences)

  • maintains respondent contact
  • solicits input for prompting
  • prompt engineering
  • analysis of the results
  • reporting and contributing to scientific publication

What we ask from you?

  • You are enrolled as MSc student
  • You have experience with the skills needed for the relevant key responsibilities listed above
  • You are able to work for 2 days a week on the project, one of which on the 11th floor of the NU VU building

Desired qualification

  • Good understanding of Dutch

What we offer

  • Close mentorship from supervisors in AI & Behaviour, the Royal Library, and the VU library
  • Hands-on experience with cutting-edge AI infrastructure
  • Opportunity to publish a scientific paper 
  • Compensation according to VU student assistant rates

Application process

Please send to Michel Klein (michel.klein@vu.nl) by October 17, 2025, end of day:

  • your CV, 
  • a brief motivation letter in which you explain how you meet the requirements
  • use as subject line: Application – RA Layman summaries

Note: This is a 16 hours/week part-time position. Employment conditions follow VU student assistant policies. The selected candidate must comply with institutional ethics and data protection guidelines.