Eye On A.I. #248 Pedro Domingos: How Connectionism Is Reshaping the Future of Machine Learning
15 snips
Apr 17, 2025 
Pedro Domingos, a prominent AI researcher and author of 'The Master Algorithm,' dives into Connectionism, the key driver of modern neural networks. He explores the historical trajectory from the 1940s to today's AI giants like ChatGPT. Discussing the tension between Connectionists and Symbolists, he illuminates the real-world impact of supervised learning and the struggles of reinforcement learning. Domingos also addresses the evolution of deep learning and the rise and fall of Generative Adversarial Networks, urging for diverse approaches in AI research.
AI Snips
Chapters
Books
Transcript
Episode notes
History of Connectionism
- Connectionism started in the 1940s by reverse engineering the human brain as the model of intelligence.
- Neural networks have had cycles of hype, decline, and revival in AI history.
Backpropagation's Central Role
- Backpropagation solves the credit assignment problem efficiently by propagating errors backward.
- It allows neural networks to adjust weights and learn complex patterns systematically.
Deep Learning to Transformers
- Deep learning breakthroughs in speech and vision paved the way for current AI progress.
- Transformers improved language understanding by enabling parallel training and flexible attention mechanisms.
