Published 12 Feb 2026
Duration: 5011
The discussion highlights the need to balance the development of highly capable AI models with efficient and cost-effective alternatives, emphasizing strategies for optimization and deployment.
From rewriting Googles search stack in the early 2000s to reviving sparse trillion-parameter models and co-designing TPUs with frontier ML research, J...
The podcast examines key areas in AI research and development, emphasizing the challenge of achieving both advanced model capabilities and efficient, cost-effective deployment. It discusses the concept of owning the Pareto Frontier by integrating hardware, model design, and techniques to develop AI systems that are both powerful and efficient. Model distillation is presented as a vital strategy for transferring knowledge from large models to smaller, more efficient counterparts, ensuring performance is maintained across different scales of deployment.
The conversation also highlights the importance of balancing theoretical innovation with practical implementation, covering topics such as the evolution of search systems, hardware advancements like TPUs, and the integration of multimodal capabilities in models like Gemini. Other considerations include the need for efficient data movement, energy-efficient computing, and the expanding ability of AI to handle complex tasks such as long-context processing, video understanding, and code generation. The discussion also touches on challenges in benchmarking, model evaluation, and system scaling, along with future goals like improving model reliability, enhancing AI reasoning, and developing more specialized hardware.
30 Mar 2026 Mistral: Voxtral TTS, Forge, Leanstral, & what's next for Mistral 4 w/ Pavan Kumar Reddy & Guillaume Lample
Mistral's Voxtral TTS is a 3B-parameter text-to-speech model leveraging neural audio codecs, semantic/acoustic token splitting, and efficient flow matching for multilingual real-time applications, balancing quality and cost while exploring future refinements in architecture, tokenization, and domain-specific training.
24 Mar 2026 Why There Is No "AlphaFold for Materials" AI for Materials Discovery with Heather Kulik
AI's role in education and materials science is debated, highlighting its potential to accelerate research in chemistry and physics through machine learning, yet acknowledging limitations in complex molecular design, data quality challenges, and the need for interdisciplinary collaboration and robust validation.
20 Mar 2026 Dreamer: the Personal Agent OS David Singleton
Dreamer is an AI platform democratizing access to agentic tools for non-technical users via customizable AI assistants, community-built apps, cross-device integration, and privacy-focused features, with a beta emphasis on accessibility, real-world productivity use cases, and third-party developer opportunities.
17 Mar 2026 Why Anthropic Thinks AI Should Have Its Own Computer Felix Rieseberg of Claude Cowork & Claude Code Desktop
Cowork is a cloud-native tool simplifying Cloud Code access for non-technical users through a balance of simplicity and flexibility, integrating with cloud services and Chrome while supporting diverse workflows via a lightweight virtual machine, developed in 10 days with a focus on modular components and ethical trade-offs in AI infrastructure.
12 Mar 2026 Retrieval After RAG: Hybrid Search, Agents, and Database Design Simon Hrup Eskildsen of Turbopuffer
TurboPuffer is a next-gen database platform focusing on AI-powered vector search, full-text search, and scalability, with a strong emphasis on hiring top talent and customer-driven innovation.