😺 🎙️ Watch: Opening AI’s black box
commentaryThe NeuronJul 16, 2026
Eric Ho of Goodfire explains why models may think in shapes, how interpretability can reduce hallucinations, and what Goodfire is building.
Impact: LowConfidence: LowNovelty: Incremental
commentarythe-neuron