Participatory Machine Learning - Rethinking the Power Dynamics

Machine Learning (ML) is gaining unprecedented momentum, it became an invisible force interweaved into our lives influencing everyday decisions. From simple applications such as movies or games recommendation systems to more complex ones such as pandemic prediction and prevention, loan illegibility and health diagnosis; ML is driving our lives!

Whilst users affected by the ML decisions are usually on the recipient end; the system designers of the so called ‘black-box’ models detain the power, they collect the data, sample it, build the ML systems, train them and more importantly interpret the outputs -for us-. The users voice is muted! Would it make difference to include the users’ perspectives? Involve the users into the development lifecycle?

On this research you will focus into using participatory research to explore ways aiming to involve the users into the design and evaluation of ML models. Would the results differ? Would this improve fairness/inclusion? Reverse the power dynamics?

Many scenarios and case studies can be envisaged for this project, we are happy to discuss other variations of the topic.

Skills Required:

A degree in Computing (or equivalent),

Artificial intelligence skills

Algorithm design and programming skills

Experience with experimental design useful

Experience with participatory research useful

Knowledge of statistics/maths useful

Background Reading:

Dean, S, Rich, S, Recht, B (2020) “Recommendations and User Agency: The Reachability of Collaboratively-Filtered Information” DOI: 10.1145/3351095.3372866

Ustun, B, Spangher, A, Liu, Y (2019) “Actionable Recourse in Linear Classification” DOI 10.1145/3287560.3287566

I Valentim, N Lourenço, N Antunes (2020) “The Impact of Data Preparation on the Fairness of Software Systems” IEEE 30th International Symposium on Software Reliability Engineering (ISSRE) Berlin, Germany, 28-31 October 2019.

Contact