Interview with Dr Hemma Philamore on decolonisation of robotics and artificial intelligence

[See bottom of page for full transcript of this video]
Mitigation for inherent bias in robotics and AI, including bias with its origin in colonialism is an important part of teaching and research because of its tangible and measurable effect on the success of robots and autonomous systems deployed in the real world. The agency of the human developer and the influence of any bias they may hold, intentional or unintentional, on the systems they work on is often overlooked. Failing to identify and act on this can negatively affect the success of robots and algorithms when deployed in the real world and can result in outcomes that discriminate against certain users.

Ongoing strategies to counter this at University of Bristol include initiatives in both research and teaching. In robotics research for example, we are engaging communities that are often overlooked in engineering in participatory research to co-author robot behaviours early on in the development process. In teaching we are introducing components of assignments that require students to identify and suggest mitigations for sources of discriminatory bias in their mathematical modelling and algorithms. Perhaps most importantly, we recognise that we are not coming from the perspective of a system that has got it right. By stimulating students to consider sources of bias in their work, our goal is to nurture the development of the next generation of engineers as individuals capable of developing solutions to current problems of discrimination in robotics and AI.