Women, Ageing and Machine Learning on Screen

Start date: 8 January 2024
End date: 31 March 2027
Funder: Leverhulme Trust
Primary investigator: Professor Joanne Garde-Hansen
External co-investigators: Dr Tanaya Guha, Glasgow University; Dr Sanjay Sharma, University of Warwick
Postdoctoral researcher: Evdoxia Taka (Glasgow University)

Partners and collaborators

BBC – Advisory Group, Learning on Screen - Advisory Group, University of Glasgow, University of Warwick

Three women smile for a selfie

Postgraduate students

Danielle Reid (University of Leeds)

Description

This project brings together new and experimental research led by arts and humanities questions: what does ageing on screen look like in UK screen cultures when AI is doing the looking? How inclusive can film and TV become if ML analyses the texts and reports back on visual ageism to industries? When ageing on screen is computed through inclusive research methods, working across disciplines of media and communication, sociology and computer sciences, how can ‘ageing on screen’ become meaningful to researchers and beneficiaries of the research?

Recent advances in ML and AI make accurate, automated analysis of complex media content more plausible than ever. More importantly, ML-based content analysis can create opportunities for deeper, nuanced interpretations and analyses of screen content that can speak to the creative sector in new ways. While the team’s previous research on automated content analysis reported a large gender gap in terms of screen time and speaking time in certain movies, we now need more insights. Everything from how many older women is on screen, for how long, are they indoors or outdoors, and when they are age estimated, what is ML showing us about the powerful social, cultural and economic frameworks created for women in culture. Our research seeks to reveal differences between ‘screen age’, ‘prosthetic age’ and ‘biological age’ to better understand cultures of ageing.

Main image caption: An example of a basic automated screen age estimation system. The age values are for illustration purposes only and are not the outputs of an ML algorithm

Impact

This research project offers a new exploration for media and screen industries of using machine learning to analyse texts at scale and the algorithm will be open source.

Publications and outputs

Pages 46-47 of Leverhulme Annual Review 2023 https://www.leverhulme.ac.uk/annual-review, https://www.flipsnack.com/leverhulmetrust/2023-annual-review/full-view.html

Pilot study at Warwick: https://warwick.ac.uk/research/priorities/connecting-cultures/blog/greydata/