Chris Olah

Canadian researcher, the figure most associated with mechanistic interpretabilitymechanistic interpretability, n.: A branch of AI research that tries to reverse-engineer a neural network — working out, piece by piece, which internal parts represent which concepts and how they combine to produce behavior. (a term credited to him), who co-founded AnthropicAnthropic, prop. n.: An American AI company that makes the Claude chatbot and models. It was founded in 2021 by former OpenAI staff and puts particular emphasis on AI safety research. in 2021 and leads its interpretability research. Without a university degree, he joined Google BrainGoogle Brain, prop. n.: Google's deep-learning research team from 2011 to 2023, where the transformer — the design behind today's chatbots — and the TensorFlow software were created., initially as an intern, co-authored the 2015 'Inceptionism' post that introduced DeepDream, and developed feature visualizationfeature visualization, n.: Making pictures that show what a part of an image-recognition network is looking for, by repeatedly adjusting an image until that part responds as strongly as possible. and attribution methods; his explanatory essays and the online journal DistillDistill, prop. n.: An online scientific journal (2016–2021) that published clear, interactive explanations of machine learning research, including founding work on understanding what happens inside neural networks. (2017), which he co-founded with Shan Carter, set new standards for visual explanation. At OpenAIOpenAI, prop. n.: The American company that makes ChatGPT and the GPT models. It began in 2015 as a nonprofit research lab and is now a for-profit company overseen by a nonprofit foundation. (2018–2020) he led work treating networks as featuresfeature (interpretability), n.: A concept or property that a neural network represents internally — such as 'is French', 'mentions a bridge' or 'this code has a bug' — usually corresponding to a direction in its internal activity. connected by circuitscircuit, n.: A small, identifiable piece of wiring inside a neural network — a few internal parts and the connections between them — that together carry out one specific job, such as spotting curves or copying a name. and proposed the universality hypothesisuniversality hypothesis, n.: The idea that different AI models trained on similar tasks tend to develop the same internal concepts and mechanisms, so what is learned about one model may carry over to others.. At Anthropic his team has published, largely in the Transformer Circuits ThreadTransformer Circuits Thread, prop. n.: Anthropic's online publication for research on how language models work inside, presented as long interactive web articles rather than conventional papers., on transformer circuits and induction headsinduction head, n.: A pattern-completing mechanism inside language models: having seen 'Harry Potter' once, it spots the next 'Harry' and predicts 'Potter' — a simple trick behind much learning from examples in a prompt., superpositionsuperposition, n.: The way a neural network squeezes more concepts into its internal activity than it has neurons, by letting them overlap — a main reason its insides are hard to read. (2022), sparse-autoencodersparse autoencoder, n.: A small add-on network trained to break an AI model's tangled internal signals into a much larger set of simpler pieces, most of which can be labeled with an understandable concept. dictionaries that extract interpretable features, first from a one-layer model (2023) and then by the million from a ClaudeClaude, prop. n.: A family of AI assistants and language models made by the company Anthropic since 2023, widely used for writing, analysis and computer programming. model (2024), and attribution graphsattribution graph, n.: A diagram showing which internal concepts led an AI model to a particular answer, with arrows showing how strongly each concept pushed the next. for circuit tracingcircuit tracing, n.: Methods that map, step by step, which internal concepts an AI model used to produce a particular answer and how they fed into one another, like tracing wires on a circuit board. (2025).