Towards Safe, Reliable and Trustworthy AI

Explore the critical aspects of developing safe, reliable, and trustworthy AI systems in this insightful conference talk by Pushmeet Kohli from Google. Delve into the rapid progress of deep learning in machine learning and artificial intelligence, and examine the challenges that arise as these models transition from research labs to real-world applications. Learn about the framework of spec-consistent machine learning and how it can help formalize notions of desired AI system behavior. Gain valuable insights into the importance of ensuring AI safety, reliability, and trustworthiness as part of the Alan Turing Institute's "Mathematics of Deep Learning" Programme. Discover how this conference brings together leading researchers and stakeholders to discuss crucial issues surrounding trustworthy artificial intelligence, including machine learning accountability, fairness, privacy, and safety. Understand the emerging directions in trustworthy AI and the engagement between academia, industry, policymakers, and the wider public in this 47-minute presentation.