Studying Machine Intelligence with Been Kim - #571
Research#AI Interpretability📝 Blog|Analyzed: Dec 29, 2025 07:42•
Published: May 9, 2022 15:59
•1 min read
•Practical AIAnalysis
This article summarizes a podcast episode from Practical AI featuring Been Kim, a research scientist at Google Brain. The episode focuses on Kim's keynote at ICLR 2022, which discussed the importance of studying AI as scientific objects, both independently and in conjunction with humans. The discussion covers the current state of interpretability in machine learning, how Gestalt principles manifest in neural networks, and Kim's perspective on framing communication with machines as a language. The article highlights the need to evolve our understanding and interaction with AI.
Key Takeaways
- •The episode discusses the need to study AI as scientific objects.
- •The focus is on interpretability in machine learning and how to improve it.
- •Communication with machines is framed as a language.
Reference / Citation
View Original"Beyond interpretability: developing a language to shape our relationships with AI"