Gary Marcus: Toward a Hybrid of Deep Learning and Symbolic AI

This podcast episode explores the current and future state of AI, examining its potential impact on human society and intelligence. Experts discuss the differences between narrow and general intelligence, the role of common sense, the limitations of deep learning, and the potential for compute to unleash new possibilities in AI. They emphasize the importance of blending symbol manipulation and deep learning approaches and the challenges of encoding tacit knowledge into AI systems. Experts also delve into the intricate relationship between innate knowledge and experience in child development and the potential for evolution to accelerate the creation of intelligent machines. The podcast concludes by highlighting the challenges in translating ethical principles into code and the pressing need to define and address harm in machine-interpretable terms.