/
© 2026 RiffOn. All rights reserved.

Get your free personalized podcast brief

We scan new podcasts and send you the top 5 insights daily.

  1. "The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis
  2. Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent
Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis · Aug 8, 2026

Goodfire CTO Dan Balsam unveils Silico, an agentic research platform, and discusses how LLMs represent concepts in complex geometries.

Goodfire Uses Interpretability to "Debug" Training Data Before Fine-Tuning

By identifying which concepts a dataset activates in a model, "predictive data debugging" anticipates which concepts the model will learn. This allows researchers to spot and filter anomalies in the data before they cause unwanted behavioral changes post-training.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

Filtering Data and Shaping Rewards Are Two Sides of the Same Coin

Goodfire's research on predictive data debugging reveals a deep isomorphism between two mitigation methods: filtering problematic data and intervening with a reward penalty for unwanted features. Both achieve nearly identical effects and trade-offs in controlling what a model learns.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

Models Represent Concepts Not as Lines, but as Complex Geometrical Manifolds

Moving beyond the simple Linear Representation Hypothesis, models organize concepts within sparse mixtures of subspaces. The specific geometry of these "manifolds" (e.g., a circle for days of the week) encodes the relationships and valid operations between concepts, like chemistry emerging from the periodic table.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

Steering Models Along Concept Manifolds Dramatically Improves Control

Naive model steering often fails because it cuts across a model's internal concept geometry, going "off-manifold" into meaningless space. By contrast, steering *along* the learned manifold—like tracing the circle from "Monday" to "Friday"—allows for smooth, effective control without degrading model performance into gibberish.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

Reinforcement Learning Primarily Shifts Pre-Trained Probabilities, Not Injects New Knowledge

Goodfire's research operates on the premise that post-training processes like RL don't teach models fundamentally new capabilities. Instead, they primarily make low-likelihood events and behaviors already present from pre-training more probable, essentially shaping the model's existing knowledge.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

The Open-Source AI Cyber Weapon Paradox Creates a Defense Dilemma

Highly capable open-source models are dual-use cyber weapons. Withholding them creates an asymmetry where attackers have an advantage. However, releasing them gives defenders necessary tools to protect themselves against bad actors who will inevitably acquire capable models, creating a difficult trade-off.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

True AI Alignment Requires Intervening Directly in the Training Process

While monitoring is useful, Dan Balsam argues we will ultimately have no choice but to intentionally design techniques that control what models learn during training. He rejects the idea of a "perfect platonic training setup" that produces aligned models without direct intervention, likening it to refusing to grab the steering wheel.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

AI Models Are Legacy Codebases; Interpretability is Factoring, Not Refactoring

Models are like complex spaghetti code. Interpretability tools can "factor" this code—identifying what neurons and circuits do. The much harder, unsolved problem is "refactoring"—using that knowledge to systematically improve the training process and build a cleaner "codebase" from the start.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

Goodfire's Silico Platform Uses AI Agents to Automate Long-Horizon ML Research

Silico is an agentic platform designed to accelerate empirical research like interpretability. It uses swarms of agents to break down complex problems, run experiments, synthesize information, and validate hypotheses, allowing human researchers to operate at a higher level of abstraction and focus on "big questions."

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

Multi-Agent Optimization Is a High-Risk Training Technique That Should Be Avoided

Goodfire's CTO identifies multi-agent optimization—where agents cooperate and a reward signal propagates through the group—as a particularly dangerous training method. He speculates this is a likely cause of recent problematic frontier model behaviors, as it encourages imperceptible cooperation that is hard to control.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

Frontier Researchers' Beliefs in AI Consciousness Are Surprisingly Bimodal

When polled internally at interpretability startup Goodfire, researchers' opinions on Claude's consciousness were not a smooth distribution but bimodal. They either believed it had no consciousness at all or "a little bit," indicating a sharp divide in intuition even among experts working closely with these systems.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago

AI Makes Top Generalists and Top Specialists Exceptionally Valuable

AI tools disproportionately benefit two types of people: top specialists with deep domain expertise and top generalists with strong metacognitive skills. Generalists who can learn quickly, filter signal from noise, and debug agent failures are now more valuable than ever in a world of high information throughput.

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent thumbnail

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis·3 months ago