Details
1090557
University of Gloucestershire
20/01/2025
6 Months
8 hours per week. We will accept alternative/flexible working arrangements e.g., working 3 days a week (over a shorter duration) for the right candidate.
Smart Casual
Pay
£16.27
£1.96
Description
Role
We are seeking a highly motivated and detail-orientated Research Assistant to contribute to an exciting research project investigating how various large language models (LLMs), like ChatGPT-4, Claude, Cohere, can be enhanced by utilising localised multimodal storytelling data to create richer and more diverse narratives while addressing sociotechnical imbalances and employing prompt engineering techniques to guide the AI in producing stories that accurately reflect local identities and experiences. This research uses a range of multimodal analysis – as such we are looking for a part-time Research Assistant with demonstrable interest and skills in Data Science research.
Project title: AI and the Art of Local Storytelling: Bridging Gaps in Identity Representation through Multimodal Narratives
About the project
This research project aims to improve large language models (LLMs) by incorporating multimodal storytelling data from local communities, allowing AI to create more authentic, culturally resonant narratives. It begins by identifying gaps in current AI storytelling, particularly in representing localized identities. The project gathers diverse narratives—visual, oral, and audio—from community-driven platforms like Gloucester Picturedrome, Voices Gloucester, and the St. Paul’s Community Mural Project, capturing unique cultural expressions and experiences. Using a multi-method approach that combines human and AI analysis, the research integrates these insights into LLMs through prompt engineering, enhancing the models’ ability to reflect nuanced local contexts. The analysis, which involves both Open-Source and Proprietary LLMs, thematically examines the content to address existing imbalances in AI storytelling. By embedding this localized knowledge, the project aims to foster more nuanced and contextually grounded narratives in AI, better aligning them with diverse community voices.
The successful research assistant will be working alongside an experienced research team drawn from across Data Science, Marketing, and Sociology subject domains. This project is funded by the School of Business, Computing and Social Science Multidisciplinary Fund.
The role
The Research Assistant will play a pivotal role in supporting the research project led by Dr Sabina Fiebig-Lord. The successful candidate will assist in data collection, analysis, and various research activities to utilise prompt engineering techniques to guide the AI in producing narratives that accurately reflect local identities and experiences and act as a technical guide on the research project.. An agile approach to development will be undertaken. The following activities and tasks are not exhaustive and may be subject to change.
Duties and responsibilities
- To work as directed on project tasks.
- Work with academic research team on the project deliverables and work package objectives. Tasks may include literature searches, literature reviews / scoping exercises, data-set collation, web site creation, administrative tasks and a potential contribution to research reports or academic publications.
- To support the delivery of the project requirements and provide regular progress updates to the line manager and project members.
- To work with colleagues in relevant roles across the University.
- To undertake any other duties and responsibilities commensurate with the grade of the post, properly directed by the line manager.
Skills and experience
. MA/MSc student in data science, AI or relevant
· Demonstrated interest and skills in extensive knowledge of LLMs and programming languages (e.g., Python). Experience using Hugging Face is a desirable requirement.
· Good understanding of the research design process
· Technical experience includes the completion of computing or data science projects. It is desirable to have experience in commercial or industry work, conducting literature searches, and delivering outputs by agreed-upon deadlines.
· Strong analytical and critical thinking skills
· Ability to work independently and as part of a team.
Location
Hybrid
Additional information
Interviews on Monday, 13th January 2025
Interested candidates should submit a 500-word cover letter outlining their interest and suitability for the position, including the names of two academic referees - either your personal tutor or relevant academic staff
Apply for job