Introduction

Responsible research and innovation depends upon active stakeholder engagement and participation as a vital means for managing the societal impact of data-driven technology. Users and stakeholders can inform and contribute to all stages of the research and innovation workflow, from ideation and design through to deployment and maintenance. However, the practice of responsible research and innovation can often be challenging due to a variety of barriers that prevent or undermine the engagement and participation of individual users or communities. 

Using specific case studies, such as crowdsourcing projects, the purpose of this group is to a) have a better understanding of the requirements, attitudes and preferences of the wider community of users, particularly minority or marginalised groups who may not have been consulted or considered, and b) to investigate how to operationalise ethical principles and best practices for supporting the co-creation of data-driven technologies (e.g. addressing barriers associated with data missingness and bias by engaging with the ‘silent’ users). 

Explaining the science

How does biased participation, as reflected in data-driven technologies and services, affect and alter our understanding of society and the environment?

Lack of inclusivity in current technologies and projects can result in the development of a technology or service that only serves the purpose of a certain group of people (i.e. leads to biased or discriminatory outcomes).

Studies of “crowd”-sourcing projects that are technically open for anyone to contribute their data to or use, including the successful "crowdsourcing" projects such as OpenStreetMap (OSM) and Wikipedia, show a significant level of bias both in terms of participants and the resulting types of data. This brings into question the scope and representativeness of the term "crowd", especially if only a small percentage of the community contributes the greatest proportion of activity (i.e. the ‘long tail effect’ or 90–9-1 rule). For example, a study has shown female mappers are more likely to represent women’s specific needs and priorities, such as hospitals, childcare services, toilets, domestic violence shelters and women’s health clinics. These are the key to driving changes in local policies, plans and budgets. So, by learning the relationships between geo-demographic features and structures at a larger scale, it is possible to trace back the structure of participants at an aggregated level and estimate the reliability of datasets for different use.

Can we identify and understand the underlying barriers to participation from missing or biased data?

A lack of certain features or data types from missing or under-representative group of participants can help us identify and understand the barriers to collecting their data and/or identify the potential ‘outliers’ for further investigations. For example, the systematic absence of specific data types or data about some regions or features can raise a flag for further studies. This missingness that has not happened at random can be studied further and improve our knowledge of the function that relates missing values and the reason for missingness. For example, the reasons why certain areas or features or values are not well recorded can tell us about the underlying reasons and barriers, which may be socio-technical in nature (e.g. the digital divide).

This has already been investigated in some cases to add a feedback loop for second iterations/versions of services. For example, lack (of granularity about) some gendered features can result in positive discrimination in favour of some gendered group in their entry or change in the process of data collection. There is evidence of male dominance in shaping OSM's tags and attributes where proposals to include spaces associated with feminised skills (e.g., ‘childcare’ or ‘hospice’) were rejected by majority/men, while sexual entertainment venues seem to have much more detail and classifications.

What are the ethical implications of addressing these scientific and sociotechnical concerns?

A core component of this project’s work will involve the careful and deliberate ethical reflection of the scientific questions and projects. This is especially important when using data to identify possible reasons for lack of representativeness in datasets, as the missingness may be the result of sensitive personal or cultural reasons (e.g. privacy concerns). Therefore, it should not be taken for granted that the pursuit of greater inclusivity and participation will be perceived as an unconditional good by all. Careful ethical reflection, dialogue, and deliberation will be embedded into each stage of this special interest group, to ensure that such issues are appropriately addressed and handled.

Aims

The primary aim of this special interest group is to understand better the needs and demands of different groups of users and stakeholders to ensure fair treatment from and access to the development of data-driven technologies, in particular for safety critical services and technologies (e.g. healthcare) or in public sector organisations that are responsible for socially significant decision-making (e.g. geolocation services). These domains are central to a wide range of Turing projects and programmes and aligned with the Turing’s goal to “advance world-class research and apply it to real-world problems”.

The group will also explore how ethical values and principles, such as responsible research and innovation, inclusivity, and participatory design, can be suitably operationalised and realised in the design and delivery of data science projects.

The wider goal is to empower the Turing community, external partners and organisations, and society more broadly, to help understand the (direct and indirect) impacts of their research or innovation on human individuals and communities.

Talking points

What are the social, technical, economic, or cultural barriers to responsible participation in data science projects?

What are the impacts and consequences of having exclusive, biased, vandalised participants in the output of data science projects?

How we can develop technologies that have diversity and inclusion by design and promote responsible and ethical participation at the heart of the project development lifecycle?

How to get involved

Click here to join us and request sign-up

Organisers

Researchers

Contact info

Christopher Burr
[email protected]

External researchers

More information coming soon.