"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis

Andreas Stuhlmüller and Jungwon Byun return to discuss how Elicit is building trusted reasoning workflows for scientific research as frontier models grow more powerful but less transparent. They explain process supervision, domain-specific reasoning primitives, and world models that make evidence, causality, and counterfactuals more inspectable. The conversation also covers life sciences use cases, evaluating conflicting evidence, automated software engineering at Elicit, token costs, Gemini, and why legible reasoning may still beat neuralese.

Mercury: Command is Mercury’s new conversational interface, giving you natural-language access to your finances and helping you take actions within your existing permissions and approval policies. Visit https://mercury.com to learn more and apply online in minutes.

LINKS:

Elicit Research Platform

Andreas Stuhlmüller Personal Site

Jungwon Byun X Profile

Ought Research Organization

Elicit Founders Previous Episode

GPT-4 Technical Report

Monitoring Reasoning Models Paper

Ought ICE GitHub Repository

Hard-to-Verify Tasks Essay

Karpathy LLM Wiki Gist

Obsidian Knowledge Base App

Mixpanel Analytics Platform

Amplitude Analytics Platform

Anthropic Tracing Thoughts Research

Claude AI Chat Assistant

METR Long Tasks Measurement

Pi Agent Scaffold Repository

Personal AI Infrastructure Repository

Elicit Claude Opus Evaluation

Elicit API Documentation

METR Developer Productivity Study

Elicit Planning Is Unsolved

Rich Sutton Bitter Lesson

Meta Llama AI Models

Recursive San Francisco Event

zero.xyz Agent Tool Access

Anthropic Dynamic Workflows Coverage

Sponsor:

Claude:

Claude by Anthropic is an AI collaborator that understands your workflow and helps you tackle research, writing, coding, and organization with deep context. Get started with Claude and explore Claude Pro at https://claude.ai/tcr

Week 2 highlights follows Anthropic’s Fable launch in real workflows, from safety gates and API refusals to autonomous coding, 3D world-building, and a Claude-run Twitter experiment. Geoffrey Irving and Daniel Murfet argue for alignment theory and guarantees before recursive self-improvement, while prinz tests Fable on legal reasoning and monitoring. Rahul Sonwalkar, Shlok Khemani, Tom McGrath, and Andrew Moore add field reports on data agents, hybrid authorship, interpretability, context systems, token economics, and power concentration.

Mercury: Run your finances with virtual cards, spending limits, merchant/category locks, and AI-friendly tools like API keys, MCP, and CLI. Check out Mercury at https://mercury.com
LINKS:

Claude Fable 5 announcement

Julius AI platform

Rahul Sonwalkar homepage

Nate Jones homepage

Shlok Khemani homepage

FrontierCode benchmark blog

Lovelace AI company

Andrew Moore Wikipedia profile

Geoffrey Irving homepage

Daniel Murfet LessWrong profile

Sequent Research announcement

Timaeus research organization

Automated Alignment paper

Goodfire AI company

Tom McGrath homepage

Predictive data debugging tool

prinzbench legal benchmark

Unit distance conjecture disproof

Dario Amodei policy essay

Vending-Bench 2 benchmark

Andon Labs site

Recursive Superintelligence startup

Sakana AI company

PostTrainBench benchmark

Thoughtful Lab company

Unit distance conjecture arXiv

Glean Work AI Index

AI Treaty open letter

Karina Nguyen homepage

Sponsor:

Claude:
Claude by Anthropic is an AI collaborator that understands your workflow and helps you tackle research, writing, coding, and organization with deep context. Get started with Claude and explore Claude Pro at https://claude.ai/tcr

Rebecca Hinds, author of "Your Best Meeting Ever" and Head of the Work AI Institute at Glean, breaks down the surprising findings from the new Work AI Index 2026 report surveying 6,000 workers. While 87% now use AI and report saving 13 hours per week, only 13% say their organization is performing significantly better—a paradox explained by two new concepts: "botsitting" (the hidden labor of making AI useful) and "botshitting" (delivering AI-generated work you can't defend). They discuss practical solutions including better-integrated AI systems, smarter AI detection policies, and aligning work to meaningful missions.

LINKS:

Rebecca Hinds Personal Website

Glean Work AI Institute

Your Best Meeting Ever Book

Glean Enterprise AI Platform

Stanford Future of Work

Glean Enterprise Graph

Pangram Labs AI Detection

OpenAI ChatGPT Product Page

Anthropic Claude Product Page

Google Gemini Product Page

Microsoft 365 Copilot Page

Glean AI Transformation 100

Rebecca Hinds LinkedIn Profile

Sponsor:

Claude:
Claude by Anthropic is an AI collaborator that understands your workflow and helps you tackle research, writing, coding, and organization with deep context. Get started with Claude and explore Claude Pro at https://claude.ai/tcr

CHAPTERS:

(00:00) About the Episode

(03:22) Grounding AI adoption

(06:46) Methodology and Glean

(12:25) Productivity paradox emerges (Part 1)

(20:31) Sponsor: Claude

(22:22) Productivity paradox emerges (Part 2)

(25:36) Bot sitting burden

(34:00) Hidden time savings

(39:56) Meaning versus automation

(47:14) Enterprise graph potential

(53:13) Detecting bot slop

(01:00:32) Retention and incentives

(01:07:54) Transformation and mission

(01:20:06) AI teammate model

(01:26:01) Future organizational design

(01:32:31) Research and meetings

(01:41:43) Episode Outro

(01:45:07) Outro

PRODUCED BY:

https://aipodcast.ing

This first highlights edition of the morning experiment tracks a week of fast-moving AI frontier news, from closed-door recursive self-improvement debates to OpenAI’s call for independent model review. You’ll hear why labs are betting on AI monitors, where safety plans still look thin, and how cheap scaffolds are already improving tax workflows. The episode also tests moderation progress and surveys AI science, cybersecurity, Vatican ethics, solo-business automation, and mental health support.

Mercury: Run your finances with virtual cards, spending limits, merchant/category locks, and AI-friendly tools like API keys, MCP, and CLI. Check out Mercury at https://mercury.com
Sponsor:

Claude:
Claude by Anthropic is an AI collaborator that understands your workflow and helps you tackle research, writing, coding, and organization with deep context. Get started with Claude and explore Claude Pro at https://claude.ai/tcr

Ali Behrouz, grad student at Cornell and Google researcher, discusses his potentially transformative work on new architectures for continual learning in AI. His paper "Nested Learning," praised by Jeff Dean as a possible paradigm shift, enables models to adapt to new context while preserving core knowledge by updating different layers at different frequencies, inspired by human memory systems. The conversation also covers his latest work on AI "sleep" for memory consolidation, why he sees all deep learning as associative memory, and the profound implications of continual learning for privacy, alignment, and the path to AGI.

Mercury: The fintech trusted by ambitious companies and individuals to run their finances, with virtual cards, spending limits, merchant/category locks, and AI-friendly tools like API keys, MCP, and CLI. Check out Mercury at mercury.com

Sponsor:

Claude:

Claude by Anthropic is an AI collaborator that understands your workflow and helps you tackle research, writing, coding, and organization with deep context. Get started with Claude and explore Claude Pro at https://claude.ai/tcr