The aim of this WP is to conduct empirical work that advances understanding of browsing and interactions with algorithm driven systems as practical activities requiring sense-making by users. Much current literature focuses on browsing behaviours such as searching, navigating, and so on as instrumental activities; however,
this is limited in scope and it is necessary to take a broader perspective to understand browsing as occurring within a situated environment in which a user employs sense-making in order to carry out different actions. This sense-making may involve drawing on implicit understandings of web platforms and technologies as well as features of the local environment. Research work to make these tacit sense-making practices visible will play an important role in assessing the occurrence and impacts of algorithm bias and in providing a foundation to move towards ‘fair’ practice in algorithm use.
In year 1 we will conduct a series of interviews and (quasi) naturalistic experiments and observations to examine user practices in relation to recommender systems, search engines and social media filters. We will identify: i) the different ways that users interact with algorithms, ii) the sense-making practices users employ interacting with algorithms, and iii) and the kinds of consequences these interactions have for browsing outcomes. Early pilot studies will inform the design of the scenarios for the ‘jury’ style workshops in WP 1 and focus groups in WP 4. Subsequent further studies will incorporate the outcomes of these ‘juries’ (WP 1) and focus groups (WP 4) and will inform the survey of existing tools (WP 2) and other activities conducted in year 2.
In year 2 we will conduct further interviews and observations alongside the hackathon and double blind comparison testing of WP 2. These activities will identify sense-making around responsibility and fair practice in interaction with algorithms. We will also conduct an ‘ethicon’ in which teams of computer scientists, engineers, ethicists and social scientists work together to write and justify algorithm design specifications. This is an innovative RRI approach which encourages participants to focus on the social and ethical impact of their designs. The ethicon will be video recorded to provide a means to capture and examine the ways in which users articulate understandings of ‘fairness’ in algorithm use. In combination the outcomes of these year 1 and 2 activities will seek to conceptualise and illustrate ‘fairness’ as practical action and accomplishment – as opposed to abstract principle.
In addition to delivering insights from action observation for use by the other WPs, The key outputs of this WP will be:
- A portfolio of instructional case studies for the training of algorithm ‘bias’ evaluation.
- A policy brief on typical user-algorithm interactions and their implication for citizen vulnerability to manipulation by automated decision making.
Academic publications on practical experiential principles of algorithm ‘fairness’.