Dublin Science Gallery illuminates bias in AI systems

In late October, the Dublin Science Gallery launched the BIAS: BUILT THIS WAY exhibition. Billed as “an interactive, thought-provoking exploration of preferences, prejudices and digital equity”, the exhibition contains works by an array of leading Irish and international artists.

An accompanying digital exhibition is showcasing a number of digital artworks that seek to explore issues such as data equity, privacy, surveillance culture, facial recognition, class and artificial intelligence.

“Can understanding human bias help build more ethical AI? Or can understanding machine bias help build more equitable societies? What can a deeper look at bias in humans and machines teach us about ourselves,” says the Dublin Science Gallery.

As AI technologies continue to become commonplace in business and government operations, attempts to interrogate how biases infiltrate these systems is vital. Once these weak spots have been found, work can begin to address them and create truly inclusive AI tools. One of the works by Karen Palmer called Perception iO (Input Output) offers visitors an immersive experience created at the “convergence of neuroscience, behavioural psychology, film, AI, facial emotion detection, eye tracking, bias and social justice.”

Taking the viewpoint of a police officer during a volatile encounter, a mounted camera tracks the eye movements and facial expressions of users, leading to a range of diverse consequences.

Another piece by Libby Heaney seeks to understand how invisible biases are translated into code. CLASSES is a video essay that takes place in a simulated London council estate. Narrated by both machine and human voices, based on research into accented speech recognition, natural language processing and public space surveillance, the work attempts to identify how historical and cultural biases around social class are being translated into code and how this affects people’s material conditions.