Skip to content
printicon
Main menu hidden.

Designing for Contestable Systems

PhD project Designing for Contestability requires new ways to navigate and negotiate the needs of various users and stakeholders. Design should address the potential power imbalance between people and predictive, decentralized systems.

This project is about the challenges involved in making it possible for people to contextualise and negotiate a data-driven system’s response to users’ actions (“response-ability”) in and through use. In this project we explore what features, mechanisms and techniques need to be designed and implemented in the front-end for users to understand, contest and possibly repair inappropriate actions by a system.

Head of project

Robert Collins
Doctoral student
E-mail
Email

Project overview

Project period:

2021-06-01 2025-06-01

Research area

Design

External funding

EU Horizon 2020 (H2020)

Project description

The need for contestability in design is becoming more apparent as we all increasingly share our world with non-human actors who have a progressively greater say in our day-to-day lives, our relationships with technology and our futures. In accepting this, we then come back to the issue of how this can be achieved with contemporary design methods (ease of use) and aesthetics (simplicity).

A solutionist approach would be to add more interface elements to reflect the workings within the ‘assemblage’ of machines, databases, algorithms, etc. which make up systems we often interact with - labels, gauges, switches, dials, graphs, buttons, etc. A specialist interface, designed for engineers by engineers to provide complete access to every stage of the process for those who have read the manual and prefer full control rather than convenience - an “expert interface”. This would appear to fly in the face of Don Norman’s tenets of UCD (Norman, D. A., 1990) and Bannon & Ehn’s intentions for Participatory Design (Bannon, L. J., & Ehn, P., 2012). The worker/user only needs to see what is necessary of the interface to do the job/task - not to hand-hold an AI/ML system through its job. This is not playing to the benefits and strengths of automated systems.

More Than Human Centred Design (MTHCD) is an approach to emerging issues within contemporary design - how to incorporate non-human entities like Artificial Intelligence, Machine Learning and other data-driven and automated decision making systems into the design process.

In an environment where decisions are being made by non-human actors, the best intentions and inherent simplifications of User Centred Design (UCD) works against and obscures the user’s ability to perceive, negotiate and potentially contest these decisions and how they are made.

There is an “...increasing rift between what computational things are and do in terms of their core character and behaviour and the ways in which they are presented to people as things for use” (Hauser et al, 2021). The ‘Black Box’ will remain opaque where the interface maintains a user centred priority.

External funding