
Machine Learning Street Talk (MLST)
Sara Hooker - Why US AI Act Compute Thresholds Are Misguided
Jul 18, 2024
Sara Hooker, VP of Research at Cohere and a leading voice in AI efficiency, shares insights on AI governance and the pitfalls of using compute thresholds, like FLOPs, as risk metrics. She critiques current US and EU policies for oversimplifying AI capabilities and emphasizes the need for a holistic view that includes data diversity. Hooker also discusses her research on 'The AI Language Gap,' revealing the complexities of creating inclusive AI that serves multilingual populations, highlighting ethical concerns and the societal implications of underrepresentation in AI development.
01:05:41
Episode guests
AI Summary
Highlights
AI Chapters
Episode notes
Podcast summary created with Snipd AI
Quick takeaways
- Using compute thresholds like FLOPS for AI governance oversimplifies capabilities and risks.
- Advocating for nuanced AI governance strategies beyond simple measures of computational power.
Deep dives
Sarah Hooker discusses multilingual AI challenges
Sarah Hooker, VP of research at Cohere for AI, explores the challenges of developing models that work across multiple languages. She highlights the limitations of existing approaches like RLHF, particularly for low-resource languages. Sarah's paper critiques using compute thresholds for AI governance, emphasizing that simple measures like FLOPS are insufficient for assessing AI capabilities and risks. She stresses the need for a nuanced approach that considers the relationship between compute, data, and model architectures.
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.