Episodes
#17 — SparksMatter inorganic design, AHOIS encoding hypothesis, RL quantum calibration
On the application of multi-agent AI systems for materials design, autonomous scientific discovery, and continuous calibration in quantum computing.
#16 — Oak Ridge plant phenotyping AI, multi-agent provenance
On using agentic AI to accelerate plant phenotyping, tracking multi-agent provenance, and managing the security and cognitive risks of GenAI integration in research.
#15 — NSF AI funding FY26, Agentic AI data problems, Robotics LLM discovery
The NSF's AI investment priorities for FY26, why agentic AI projects are stalling, and how LLMs are being used for autonomous scientific discovery.
#14 — OpenAI GPT-5.4 boosts drug synthesis, DN-Hypo-Pipeline generates hypotheses
On new AI systems for scientific discovery, including workflows for hypothesis generation, robotic chemistry, and agent-based systems for math and protein prediction.
#13 — LifeSciBench, NeuralTrust AI security funding, ProteinGym prediction gain
An overview of new developments in AI for science, including the LifeSciBench benchmark and new methods for autonomous research agents.
#12 — DeepMind AI Control Roadmap, OpenAI drug discovery, AI persona authorship
An AI Control Roadmap from Google DeepMind proposes managing advanced AI as an internal security threat, one of several recent developments in AI research and safety.
#11 — Singapore AI-for-Science, Omnimodal agent orchestration framework
On recent AI developments, including Singapore's national science initiative, updated UK and US governance proposals, and new research on agentic AI and LLM moral judgment.
#10 — Orion automates biomedical image, data analysis
On Orion, an AI agent for biomedical image analysis that automates computational workflows by combining large language models with terminal and GUI control.
#9 — NIH Bridge2AI, Mistral Physics AI, LLMs Outperform Medical AI
On the NIH's new Bridge2AI program for biomedical research, Mistral AI's move into "physics AI", and general-purpose LLMs outperforming specialized medical AI.
#8 — AI agents shift to harness engineering, scientific AI workflow struggles
On the development of AI agents prioritizing 'harness engineering' for reliability and the current challenges facing AI-powered scientific workflow automation.
#7 — EurekAgent scientific discovery, Evo autoresearch, LLMs direct chemistry
An overview of new AI systems like EurekAgent for automating scientific research, and frameworks for ensuring AI safety and trustworthiness.
#6 — Anthropic Mythos 5, Fable 5, Canada $2.3B AI strategy
On new AI models for science, methods for research automation like 'loop engineering' and multi-agent systems, and recent government AI policy.
#5 — DeepMind funds AI safety, Cornell AI safety, DARPA decentralizes AI
An overview of new funding, research initiatives, and frameworks for developing and managing safe, coordinated multi-agent AI systems for science and enterprise.
#4 — The Rise of AI Agents in Scientific Discovery
An overview of how agentic AI systems are accelerating scientific discovery, the challenges they present, and the continuing need for human expertise in research.
#3 — Multi-Agent AI Systems in Scientific Research
On the use of multi-agent AI systems to automate scientific discovery, and the design principles and verification protocols required for their reliable operation.
#2 — AI Agents in Scientific Research
A look at how autonomous AI agents are automating complex workflows in scientific research, from programming and chemistry to genomics and microscopy.
#1 — AI Tools for Accelerating Scientific Research
An overview of new AI systems from Google, Microsoft, and FutureHouse designed to accelerate scientific research in fields from mathematics to quantum computing.