LLMs as a Service, Cohere for AI Research Vision, and Trustworthy AI — Interview with Sara Hooker

Sara Hooker leads Cohere for AI, a non-profit research lab to solve complex Machine Learning problems. Sara has previously worked as a research scientist at Google Brain focusing on going beyond test-set accuracy to evaluate models that need to fulfill multiple desired criteria such as interpretability, fairness or robustness.

In the interview, we dive into the nature of Cohere For AI, her famous paper "The Hardware Lottery", challenges in building and deploying trustworthy AI...👇

📚 Related resources:
Cohere For AI: https://cohere.for.ai
Hardware Lottery Paper: https://arxiv.org/abs/2009.06489
Sara Hooker homepage: http://www.sarahooker.me
Learn more about Zeta Alpha: https://www.zeta-alpha.com


⏱ Timestamps:
00:00 Intro
00:48 Sara's professional journey
05:00 Research topics at Cohere For AI
06:50 The hardware lottery and Transformers
07:30 Large Language Models as a service, what's the future?
12:39 Multimodal Foundation Models
14:49 On Deep Learning Efficiency
20:04 Trustworthy AI main challenges
23:54 Minimizing harm from publicly released powerful models
27:21 How do you keep up with research at Cohere For AI? Specialists vs. Generalists