Published 27 Feb 2026
Duration: 3374
The podcast explores the intersection of AI capabilities and societal risks, discussing challenges in evaluating and deploying AI systems.
This is a free preview of a paid episode. To hear more, visit www.latent.spaceAIE Europe CFP and AIE Worlds Fair paper submissions for CAIS peer revie...
The podcast explores the META framework, a method for evaluating AI models through two main components: Model Evaluation (M-E), which assesses AI capabilities and real-world performance, and Threat Research (T-R), which investigates potential societal risks associated with AI advancements. It introduces the Model Time Horizon Chart, a visualization that depicts the linear progression of model capabilities over time, based on the difficulty and reliability of human-equivalent tasks. The discussion also covers challenges in measuring AI performance, the selection of appropriate tasks for evaluation, and how AI advancements could impact developer workflows and productivity.
The conversation further delves into concerns related to AI safety, the potential for rapid capability growth, and the future of AI autonomy. It addresses limitations in current benchmarking systems, the role of increasing computational power, and the possibility of AI systems improving themselves independently. The podcast also touches on topics like AI-driven trading and prediction markets, along with the ethical considerations that arise from evaluating and deploying AI technologies. Finally, it reflects on the ongoing evolution of AI research and the importance of developing open-ended evaluation methods to balance progress with long-term safety and risk management.
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.