Exposing Bias in AI-Driven Systems: AI Researchers and Artists launch BIAS Exhibition at Science Gallery Dublin

22 October 2021

AI-driven systems are subject to the biases of their human creators who unintentionally embed biases into systems either by training them on biased data or ‘rules’ created by humans with implicit biases. Understanding how bias can affect systems and breaking the cycle of bias with the development of trustworthy AI has the potential to promote greater equality across contexts — from business to science to the arts — on dimensions including gender, race, socioeconomic status, and others.  Researchers from the ADAPT Centre for AI-Driven Digital Content Technology have partnered with artists to co-create a new exhibition at Science Gallery Dublin, ‘BIAS: BUILT THIS WAY’, opening today (22 October 2021).

The exhibition, which opens today, challenges visitors to look at our own biases in everyday life. With our growing reliance on AI-driven technologies to become the decision maker in high-stakes human contexts such as hiring, access to credit, insurance and more, BIAS looks to interrogate how prejudice can creep into the training data of an AI system which can then influence future decisions and embed the bias. This brand-new exhibition interrogates fairness, the ethics of Artificial Intelligence, machine learning and data processing in humans and algorithms.

As a part of this exhibition commissioned by Science Gallery Dublin, ADAPT has four artist researcher collaborations:

Flag for Removal

What’s it like to be a human worker in a world designed for algorithms?

Social media companies process billions of posts on any given day. They use AI to automate the screening and moderation of this material, but still rely on human moderators to validate and ensure the accuracy of the algorithm. This approach seems to offer the best of both worlds – the speed of AI and the eye of context and meaning of a human – but what is the experience of the human worker in a system designed to function at the speed of AI? This interactive game draws attention to the high-pressure working environment many content moderators experience on the job.

Created by: Aphra Kerr (ADAPT Maynooth), Caroline Sinders & Karl Hohn. Research assistance was provided by Melina Garcia, Ciara Hogan and Joi Neuda.

Your Balanced Media Diet

Is awareness of your media consumption the first step in combating it? 

Your Balanced Media Diet is a playful interactive installation that challenges visitors to reconsider the new sources they tend to consume without question. Just as they would their food, the visitor is invited to examine their intake and learn how they can develop a balanced news diet. Visitors can assess the bias in their media consumption and explore the NFC empowered food items in the fridge to explore alternatives.

Created by: Brendan Spillane (ADAPT TCD), Ross Dowd

The Uncanny Valley of Breath

What does a breath mean to a machine?

When we talk to a machine, we don’t provide it with polished packets of information that it can analyse. Instead, we send it a chaotic bundle of vibrating air that must be searched for recognizable patterns. For an AI trained in Natural Language Processing, the way humans breathe when talking is useless – an undecipherable complexity. The Uncanny Valley of Breath strips away all of the processing that machines do to show us the variety in human speech in its most basic form: as a visualisation of frequencies, extracted from our speech, projected onto walls of spinning fans simulating the airflow that generated them.

Created by: Cormac Patrick English (ADAPT UCD), Edward Storey (ADAPT TCD), Antony Nevin, Carlotta Aoun.

SKU-Market

Who does an algorithm think you are?

This exhibit explores how purchasing habits and online behaviours can be interpreted, skewed, and applied to shape our lives in unexpected ways. The Algorithmic Aisle shelves are lined with products you’ll love. Shop for what you like and then visit Self-Checkout. SKU-Market’s proprietary algorithm will build a vivid profile just for you. But watch out: you might not like who SKU-Market thinks you are. 

Created by: Jennifer Edmond (ADAPT TCD), Laura Allcorn, Ivan Bacher (ADAPT). 

The ‘BIAS: BUILT THIS WAY’ exhibition opens on Friday 22nd of October and will run until early 2022. BIAS is part of a year-long programme of events at Science Gallery Dublin exploring AI, Ethics, Trust and Justice. The programme focuses on two kinds of activity – incubation and activation – through exhibitions, events and education, questioning the social, psychological and technological aspects of bias.  More information online at https://dublin.sciencegallery.com/bias