Using Generative AI to explore social media's impact on adolescent health

Summary

Adolescents are the most active age group on social media, spending just under 5 hours a day online. They have also been experiencing a pronounced drop in mental health over the past decade.

Concerns that social media is partially causing this decrease have spread to the highest levels of international decision making, such as the UK House of Commons Education Select Committee and the US Surgeon General (“there is an urgent need for additional research including on…the potential pathways through which social media may cause harm to children’s and adolescents’ mental health”). With mental health problems already costing ~£100 billion annually in England alone, understanding whether social media might be adding to this burden is time critical.

There are many potential mechanisms through which social media could be impacting adolescent health, including through the content consumed. However, it has until now been largely impossible to understand the impact of content consumption on a large-scale level as both the data and methods have been unavailable.

There is urgent need for progress. Data infrastructure innovations and the advent of generative AI have led us to the precipice of now being able to deliver high-intensity objective content consumption data from young people linked with longitudinal health outcomes. This project pioneers such a novel approach to deliver a step-change to our understanding of how social media content impacts adolescents’ health. It will set the foundations for further research across a range of social media content types and individual outcomes.

Project aims

The project will analyse pre-existing and newly collected data which includes logs of TikTok videos viewed by young people and ecological momentary assessments of their mental health and other outcomes of interest.

You will develop methodological pipelines for using Generative AI to code this large-scale objective content consumption data from TikTok, using their API.

You will then apply these methods to our data to test how consumed relates to mental health on a between-person and within-person level, gaining expertise in complex longitudinal modelling. 

You will be encouraged to engage consistently with young people and policymakers throughout the PhD to ensure the work is informed by stakeholders and has direct national impact. 

Contact details

Dr Amy Orben - amy.orben@mrc-cbu.cam.ac.uk

Opportunities

This project is open to applicants who want to do a:

• PhD