Machine Learning Limitations
Gary argues that deep learning excels in object classification but struggles with abstract knowledge representation. He suggests that while neural networks may form high-level abstractions, they often remain inaccessible to humans. The conversation highlights the need for new approaches to machine learning that can better capture the complexities of our world.In this clip
From this podcast

Lex Fridman Podcast
Gary Marcus: Toward a Hybrid of Deep Learning and Symbolic AI | Lex Fridman Podcast #43
Related Questions