Celeste Kidd at AI Debate 2
BCNM faculty member Celeste Kidd featured as a speaker at MONTREAL.AI's second AI Debate titled "Moving AI Forward: An Interdisciplinary Approach". The debate consisted of three panels: "Architecture and Challenges", "Insights from Neuroscience and Psychology" and "Towards AI we Can Trust". Alongside Ryan Calo, Margaret Mitchell and Francesca Rossi, Celeste participated in Panel 3: "Towards AI we Can Trust".
In her remarks "Profound impacts of AI (and its biases) on human beliefs", Celeste explains why algorithmic bias is problematic for both the direct harm it causes and the cascading harm it can bring to human beliefs. From her presentation:
This is problematic because there are systems interfacing with people everyday, embedded seamlessly within our lives. These systems drive human beliefs in sometimes destructive and likely irreparable ways. Our research shows that
(1) People don't learn deeply about most things.
(2) People make up their minds quickly.
(3) Once a person has made up their mind about something, they don't easily revise.
Biases in AI systems reinforce and strengthen biases in the people who use them. The termination of Timnit Gebru marks a dark term. It's clear that private interests will not support diversity, equity, and inclusion. It should horrify us that the control of algorithims that control so much of our lives lie in the hands of a homogenous, narrow-minded minority.