Weaviate Podcast #8 • Brady Neal about Causal Inference in Vector Search
Brady Neal from Oogway talks with Connor Shorten from Henry AI Labs about causal inference and many more. See the timestamps below to check out what this podcast is all about.
Oogway Website: https://www.oogway.ai/
Weaviate docs: https://link.semi.technology/3xy57fi
Embeddings API in Weaviate: https://link.semi.technology/3Jjk5LJ
Weaviate Github: https://link.semi.technology/3EtTlGn
Weaviate slack: https://link.semi.technology/3d3x7iU
Weaviate newsletter: https://link.semi.technology/2ZS9ypI
Weaviate: https://weaviate.io/
0:00 Introduction
0:40 What is Causal Inference?
1:40 Disentangled Representations and CI
3:25 Modeling unobserved factors in Machine Learning models
5:00 Causal Language Modeling
13:00 Causal Inference versus Reinforcement Learning
20:38 Neurosymbolic AI
22:15 Vectors to Causal Pathways
26:25 Conditional Probability versus Causal Relationships
31:45 Robustness to Negations
35:20 Causal Structure Discovery for Neural Network Interpretability
38:20 Oogway AI
41:30 Architecture of combining AI services in an “App store” style design
47:40 Mixture of Experts Routing across Search APIs
54:40 Human-Computer Interaction
1:03:00 Thoughts on Weaviate, Jina, and Haystack
1:04:45 Decentralized Startups
1:16:30 Advice on Information Diet