Anthropic's work on reading a model's internal "thoughts" is more than a safety feature; it's a new frontier for performance. The ability to "train the thoughts, not just the words" gives developers a direct lever to improve a model's internal reasoning, fix failures, and enhance reliability, moving interpretability from theory to practice.

Mechanistic interpretability (Mekinterp) research has been slow due to its manual, ad-hoc nature. The guests argue that coding agents can automate the experimentation process, enabling large-scale, systematic analysis of AI models. The first science AI should automate is the science of understanding itself.

Goodfire frames interpretability as the core of the AI-human interface. One direction is intentional design, allowing human control. The other, especially with superhuman scientific models, is extracting novel knowledge (e.g., new Alzheimer's biomarkers) that the AI discovers.

Attempting to interpret every learned circuit in a complex neural network is a futile effort. True understanding comes from describing the system's foundational elements: its architecture, learning rule, loss functions, and the data it was trained on. The emergent complexity is a result of this process.

The ambition to fully reverse-engineer AI models into simple, understandable components is proving unrealistic as their internal workings are messy and complex. Its practical value is less about achieving guarantees and more about coarse-grained analysis, such as identifying when specific high-level capabilities are being used.

The field is moving beyond labeling concepts with sparse autoencoders. The new frontier is understanding the intricate geometric structures (manifolds) these concepts form in a model's latent space and how circuits transform them, providing a more unified, dynamic view.

Just as biology deciphers the complex systems created by evolution, mechanistic interpretability seeks to understand the "how" inside neural networks. Instead of treating models as black boxes, it examines their internal parameters and activations to reverse-engineer how they work, moving beyond just measuring their external behavior.

We don't fully understand how advanced AI models work. Creators don't program them with explicit knowledge but train them on vast datasets and then run experiments to discover their capabilities. This makes AI development more of a science—studying an unpredictable artifact—than traditional engineering, highlighting an inherent lack of control.

By analyzing a model predicting Alzheimer's, Goodfire discovered it relied on the length of cell-free DNA fragments—a previously overlooked signal. This demonstrates how interpretability can extract new, testable scientific hypotheses from high-performing "black box" models.