How can the BBC employ artificial intelligence to deliver data-driven personalised media content recommendations responsibly?

The Midlands3Cities Doctoral Training Partnership is offering four fully-funded PhD studentships to work on innovative projects in the areas of Artificial Intelligence and Data-Driven Research. M3C DTP is a collaboration between six leading universities in the Midlands and successful students will join the wider M3C doctoral community.

Each research project is a collaboration between a M3C university and an external partner. Further details of the University of Birmingham project can be found below:

Cambridge Analytica’s bulk-harvesting of Facebook profiles for political micro-targeting highlights how data driven personalisation of media content could be employed to manipulate individuals at scale, undermining democracy. Yet using AI to personalise information services generates substantial user benefits, tailoring content to users’ interests, delivered continuously and conveniently while helping users manage information overload. These well-established practices have potentially profound consequences, yet their legal and ethical risks are poorly understood, requiring urgent attention.

These risks are particularly acute for the BBC, because such practices directly implicate its core duty, as the nation’s public service broadcaster, “to act in the public interest, serving all audiences through the provision of impartial, high-quality and distinctive output and services which inform, educate and entertain.”

This project will investigate how the BBC can utilise AI to deliver personalised media content recommendations (‘PMCR’) responsibly, examining how to do so lawfully, ethically, value-sensitively and practically.

An interdisciplinary, mixed methods approach will be adopted, utilising multiple sources including (a) normative analysis of primary and secondary legal and ethical sources ; (b) technical analysis of algorithmic methods and approaches to content recommendation; (c) fieldwork and site visits to the BBC, including interviews and surveys with key staff including attendance at BBC’s Machine Learning Ethics committee meetings; (d) analysis of historical publicly available data (e) field experiments and surveys of BBC user attitudes; and (e) use of ICT to provide a human-centric and BBC Charter-compliant approach to responsible PCMR.

The studentship will normally be for up to three years for full-time study, or up to six years for part-time study, starting at the beginning of the 2018/19 the academic year.