Gabe Gomes, Assistant Professor at Carnegie Mellon University, discusses Democratizing Scientific Knowledge using AI. They talk about Coscientist, the first non-organic, intelligent being to design, plan, and execute a chemistry experiment. They explore the benefits and concerns of using AI in science, the potential of AI in generating synthesis protocols, the value of automation in scientific research, and the democratization of scientific knowledge using AI.

Dan O'Connell, Chief AI & Strategy Officer at Dialpad, discusses building language models, custom speech recognition, and the challenges of AI in business communication. They explore the potential of AI in sales and customer service, evaluating AI performance, and the dynamics of fine-tuning work in AI.

Shane Legg, Cofounder at Deepmind, shares insights on AGI's timeline, new architectures, and the future of multimodal AI. They discuss measuring progress towards AGI, aligning AI models ethically, scaling systems, Deepmind's influence, and the limitations of current models.

Neurotechnology expert Nita Farahany discusses the current state of neurotechnology, including devices like SmartCap and Neuralink, cognitive liberty, risks and need for regulation. They also explore the broader applications of neurotechnology in national security, controlling drones with the mind, brain-to-brain communication, and the potential impact on work, healthcare, privacy, and human cognition.

Samuel Hammond, economist for the Foundation for American Innovation, discusses AGI's impact on economies, institutions, and power structures. They explore the default path to AGI, economic transformations, and AI safety megaprojects. The conversation also covers in-context learning, AI progress in relation to evolution, occupational licensing, transaction costs, euphemisms, Estonia's IT advancements, dangers of AI implementation, alignment in AI models, and the need for government intervention in AI.

Topics discussed include the need for more AI scouts, OpenAI's safety record, redteaming frontier models, the firing and rehiring of OpenAI's CEO, the importance of curiosity and understanding in AI, learning from Masterclass, an insider perspective on OpenAI, events at OpenAI and the polarization of AI discourse, early access to OpenAI's GPD4, reflections on initial conversations, the capabilities and limitations of GPD4, the need for control and safety in AI behavior, advancements and challenges of Chat GPT, questioning the narrative on China and AI, the simplicity and power of AI systems, AI in the medical field and the debate around AI technology, disagreement and breakdown in communication within the board, knowledge sharing and collaboration at OpenAI, the release of Meta, a powerful AI model with ethical concerns, and recommendations for listeners on how to keep up with the podcast.

This emergency pod delves into Mamba, a new state space model architecture. It explores the disparities between human cognition and AI, the limitations of current language models, and optimizing performance in linear algebra computations. It also discusses the importance of high-level narrative and reasoning in training AI and the challenges of interpretability and safety in AI models. The podcast ends by exploring the potential of a new era in AI with specialized architectures and advancements in AI agents.

Nathan chats with Dan Shipper, CEO and Co-founder of Every, about using ChatGPT as a copilot for creative and cognitive labor. They discuss Nathan's prompting techniques, using GPT in copilot mode, and struggles in content creation. They also explore best practices for reasoning, using ChatGPT as a copilot for a React app, and building renderers inside ChatGPT. The speakers express excitement for the future and invite listeners to join as hosts or sponsors.

Marius Hobbhahn, CEO of Apollo Research, discusses AI deception, interpretability, and affordances. They explore the behavior of AI systems, the importance of auditing AI models, and the limitations of AI. They also discuss theory of mind in AI systems, deceptive behavior in AI models, and the need for third-party auditing and red teaming.

Balaji Srinivasan, former CTO at Coinbase and investor, discusses polytheistic AI, human-AI symbiosis, and prospects for AI control. They explore the potential power and limitations of AI, challenges in ensuring robustness, offense and defense in technology, decentralized AI, the potential path to an AI disaster, cryptographic control, shifts in the American workforce, and the polarization of AI discourse.