The Index Podcast
AI Governance and Decentralized Training with Alexander Joseph Long, Founder of Pluralis.ai
Sep 6, 2024
Alexander Joseph Long, the founder of Pluralis AI and an expert in non-parametric external memory in deep learning, dives into the future of AI governance. He discusses the importance of decentralized training for ownership and control of AI models. As Ethereum shifts from proof of work to proof of stake, he highlights exciting possibilities for AI using repurposed GPUs. The conversation also addresses the challenges of centralization in AI education and the urgent need for innovative regulatory solutions in the fast-evolving tech landscape.
37:37
Episode guests
AI Summary
AI Chapters
Episode notes
Podcast summary created with Snipd AI
Quick takeaways
- Decentralized training of AI is essential for promoting ownership and control, challenging the monopolistic power of major tech companies.
- Practical challenges in decentralized AI training, including low bandwidth and latency, require innovative solutions to enhance communication efficiency.
Deep dives
Decentralized Training and Its Significance
Decentralized training of AI is pivotal for ensuring ownership and control over models. When a model is created by a centralized group, such as a major tech company, they maintain the power to dictate how the model behaves and its usage. This centralization presents significant risks, particularly in creating a landscape where a small number of entities control the technology that influences societal decisions. By promoting decentralized training, the vision is to foster a more democratic governance model over AI, ensuring that the models are accessible and modifiable by a broader community.
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
