Published Feb 3, 2022

Connor Leahy on EleutherAI, Replicating GPT-2/GPT-3, AI Risk and Alignment

Connor Leahy delves into the formation and growth of EleutherAI, the challenges of replicating GPT-2/GPT-3, and the critical importance of AI alignment to mitigate existential risks, offering deep insights into the world of open-source AI research and community dynamics.

Episode Highlights

Topics covered

Episode Highlights

Related Episodes

Jeremie Harris: Realistic Alignment and AI Policy
Answers 383 questions
Percy Liang on Machine Learning Robustness, Foundation Models, and Reproducibility
Answers 383 questions
Kyunghyun Cho: Neural Machine Translation, Language, and Doing Good Science
Answers 383 questions
Sara Hooker: Cohere For AI, the Hardware Lottery, and DL Tradeoffs
Answers 383 questions
Ed Grefenstette: Language, Semantics, Cohere
Answers 383 questions
Evan Hubinger on Effective Altruism and AI Safety
Answers 383 questions
Yann LeCun on his Start in Research and Self-Supervised Learning
Answers 383 questions
Jeff Clune: Genetic Algorithms, Quality-Diversity, Curiosity
Answers 383 questions
Vera Liao: AI Explainability and Transparency
Answers 383 questions
Daniel Situnayake: AI on the Edge
Answers 383 questions
Scott Aaronson: Against AI Doomerism
Answers 383 questions
Seth Lazar: Normative Philosophy of Computing
Answers 383 questions
Irene Solaiman: AI Policy and Social Impact
Answers 383 questions
Peter Lee: Computing Theory and Practice, and GPT-4's Impact
Answers 383 questions
Nicholas Thompson: AI and Journalism
Answers 383 questions

Dexa/Daniel Bashir

Connor Leahy on EleutherAI, Replicating GPT-2/GPT-3, AI Risk and Alignment

Topics covered

Popular Clips

Small Timelines, High Outcomes

Safety Concerns

The Power of GPT-2

Uncertainty and Regulation

Exploring AI Origins

Nuances in AI Ethics

Unconventional Essay Writing

The Future of AI

AI Alignment Concerns

The Birth of Luther

Shifting Perspectives

Alpha AI Community

AGI and Neural Networks

Releasing Large Language Models

Learning from Isolation

Episode Highlights

AI Risk and Alignment

AI Origin Stories

Community Building

Related Episodes

Jeremie Harris: Realistic Alignment and AI Policy

Percy Liang on Machine Learning Robustness, Foundation Models, and Reproducibility

Kyunghyun Cho: Neural Machine Translation, Language, and Doing Good Science

Sara Hooker: Cohere For AI, the Hardware Lottery, and DL Tradeoffs

Ed Grefenstette: Language, Semantics, Cohere

Evan Hubinger on Effective Altruism and AI Safety

Yann LeCun on his Start in Research and Self-Supervised Learning

Jeff Clune: Genetic Algorithms, Quality-Diversity, Curiosity

Vera Liao: AI Explainability and Transparency

Daniel Situnayake: AI on the Edge

Scott Aaronson: Against AI Doomerism

Seth Lazar: Normative Philosophy of Computing

Irene Solaiman: AI Policy and Social Impact

Peter Lee: Computing Theory and Practice, and GPT-4's Impact

Nicholas Thompson: AI and Journalism

Connor Leahy on EleutherAI, Replicating GPT-2/GPT-3, AI Risk and Alignment

Topics covered

Popular Clips

Episode Highlights

AI Risk and AlignmentConnor Leahy discusses the existential risks of advanced AI, emphasizing the importance of alignment to prevent unintended consequences. He explores the challenges of AI alignment research and the ethical considerations of releasing large language models.

AI Risk and Alignment

AI Origin StoriesConnor Leahy, co-founder of EleutherAI, shares his journey from a science enthusiast to an AI researcher focused on alignment and open-source projects. His story highlights the unexpected opportunities and challenges he faced while replicating GPT-2 and exploring AI alignment.

AI Origin Stories

Community BuildingConnor Leahy shares the story of EleutherAI's formation, highlighting the community's grassroots origins and its focus on replicating GPT-3. He discusses the dynamics of maintaining an online research community and the challenges of sustaining initial excitement.

Community Building

Related Episodes