Bias in AI and Neuroscience Transdisciplinary conference
This conference intends to bring together scholars from multiple disciplines, including cognitive sciences, social sciences, and humanities, to discuss the challenges that bias brings to the production and the application of scientific knowledge in neuroscience and AI and practical strategies and methodologies to mitigate these challenges.
Neuroscience and AI are playing a powerful role in our understanding of, and dealings with, human differences. For example, brain imaging is widely treated as a privileged source of evidence for the veracity or naturalness of various kinds, including identity markers such as sexual orientation. Brain imaging and machine learning algorithms are also advancing the field of precision medicine, a context in which efforts to stratify subjects according to sex and race have taken on renewed significance. Furthermore, as algorithms are impacting more and more areas of daily life, some hope that AI will enable us to eliminate bias from all processes of categorization and prediction (including hiring practices and parole decisions). At the same time, troubling technological blind-spots have been revealed by a number of incidents – ranging from voice-activated car technology unable to recognize women’s voices to image recognition software unable to correctly classify non-white faces.
Such technological and scientific developments warrant critical consideration of the ways in which differences and categories are made (and perhaps un-made) in neuroscience and AI, and of the ways in which bias affects or emerges from this. The time is ripe to have open and ethically informed discussions on what is needed to navigate these challenges.