Srishti Gautam, a PhD candidate at UiT Machine Learning Group, provided a presentation titled "Trustworthy and Fair AI" (18 Jan. 2024)
Abstract:
Artificial intelligence (AI) has seen significant advancements with the creation of sophisticated deep learning models that excel in multiple areas. Nonetheless, these advancements come with pressing challenges, as deep learning models may reflect and intensify the biases in their training data. Moreover, the complexity of these models leads to a lack of transparency, potentially hiding these biases and eroding trust, hindering their wider acceptance. It is therefore crucial to foster the creation of AI systems that are inherently transparent, trustworthy, and fair. This talk presents a universal method capable of converting any existing pre-trained black-box model into a self-explainable one, thereby addressing the issue of lack of transparency. Further, the discussion will also delve into the widely used Large Language Models, exposing their embedded unfairness and the perpetuation of social biases.