An AI with an Agenda - How Our Cognitive Biases Leak Into Machine Learning
Explore the hidden biases in AI and machine learning systems in this thought-provoking conference talk. Delve into how human cognitive biases can inadvertently seep into algorithmic decision-making processes, challenging the notion of AI's objectivity. Learn about real-world consequences of biased AI, from mortgage lending to predictive policing. Discover practical strategies to mitigate these biases, including the importance of asking the right questions, using conditional probability, building representative datasets, and implementing distributionally robust optimization. Examine case studies like Amazon's hypothetical salary tool and the "bratwurst problem" to understand the complexities of AI bias. Gain insights into ethical considerations, transparency in AI use and algorithms, and the role of control engineering in creating fairer AI systems. Walk away with a deeper understanding of how to develop more equitable and reliable AI and machine learning models that can truly benefit society. Read more