Published Jul 18, 2024

Sara Hooker - Why US AI Act Compute Thresholds Are Misguided

Sara Hooker delves into the ethical dilemmas and computational challenges in AI, critiquing U.S. AI Act compute thresholds, and advocating for nuanced policy measures while addressing biases in multilingual language models.

Episode Highlights

Topics covered

Episode Highlights

Related Episodes

#92 - SARA HOOKER - Fairness, Interpretability, Language Models
Answers 383 questions
Sara Hooker - The Hardware Lottery, Sparsity and Fairness
Answers 383 questions
Nora Belrose - AI Development, Safety, and Meaning
Answers 383 questions
#88 Dr. WALID SABA - Why machines will never rule the world [UNPLUGGED]
Answers 383 questions
How Do AI Models Actually Think? - Laura Ruis
Answers 383 questions
#57 - Prof. Melanie Mitchell - Why AI is harder than we think
Answers 383 questions
CAN MACHINES REPLACE US? (AI vs Humanity) - Maria Santacaterina
Answers 383 questions
Taming Silicon Valley - Prof. Gary Marcus
Answers 383 questions
#046 The Great ML Stagnation (Mark Saroufim and Dr. Mathew Salvaris)
Answers 383 questions
Jurgen Schmidhuber on Humans co-existing with AIs
Answers 383 questions
#56 - Dr. Walid Saba, Gadi Singer, Prof. J. Mark Bishop (Panel discussion)
Answers 383 questions
AI Alignment & AGI Fire Alarm - Connor Leahy
Answers 383 questions
Sayash Kapoor - How seriously should we take AI X-risk? (ICML 1/13)
Answers 383 questions
#034 Eray Özkural- AGI, Simulations & Safety
Answers 383 questions
Joscha Bach and Connor Leahy on AI risk
Answers 383 questions

Dexa/Machine Learning Street Talk (MLST)

Sara Hooker - Why US AI Act Compute Thresholds Are Misguided

Topics covered

Popular Clips

Dynamic Learning Approaches

Ethical Frameworks in ML

Hardware and Algorithm Evolution

Machine Learning Insights

Language Model Optimization

Algorithmic Priorities

Language Game Insights

Steerability and Creativity

Language Model Challenges

Model Lifecycle Challenges

Risky Model Threshold

Neural Creativity Unleashed

Compute and Capabilities

Machine Learning Insights

Episode Highlights

Ethical Considerations in AI

Compute Threshold Debate

Multilingual AI Challenges

Related Episodes

#92 - SARA HOOKER - Fairness, Interpretability, Language Models

Sara Hooker - The Hardware Lottery, Sparsity and Fairness

Nora Belrose - AI Development, Safety, and Meaning

#88 Dr. WALID SABA - Why machines will never rule the world [UNPLUGGED]

How Do AI Models Actually Think? - Laura Ruis

#57 - Prof. Melanie Mitchell - Why AI is harder than we think

CAN MACHINES REPLACE US? (AI vs Humanity) - Maria Santacaterina

Taming Silicon Valley - Prof. Gary Marcus

#046 The Great ML Stagnation (Mark Saroufim and Dr. Mathew Salvaris)

Jurgen Schmidhuber on Humans co-existing with AIs

#56 - Dr. Walid Saba, Gadi Singer, Prof. J. Mark Bishop (Panel discussion)

AI Alignment & AGI Fire Alarm - Connor Leahy

Sayash Kapoor - How seriously should we take AI X-risk? (ICML 1/13)

#034 Eray Özkural- AGI, Simulations & Safety

Joscha Bach and Connor Leahy on AI risk

Sara Hooker - Why US AI Act Compute Thresholds Are Misguided

Topics covered

Popular Clips

Episode Highlights

Ethical Considerations in AI

Compute Threshold DebateSara Hooker, VP of Research at Cohere, critiques the use of compute thresholds in AI governance, highlighting their limitations in assessing AI risk. She discusses alternative measures and historical lessons that could inform more effective policies.

Compute Threshold Debate

Multilingual AI ChallengesSara Hooker explores the biases in language models due to uneven language support, highlighting the impact on model performance and cultural representation. She discusses data inequality and innovative optimization strategies to improve multilingual AI models.

Multilingual AI Challenges

Related Episodes