AI research, mechanistic interpretability, and philosophy exploring how LLMs work and what that reveals about the algorithms behind both machines and humans.

Layers of Meaning: How Detokenization and Abstraction Impact Prompt Design

What Prompting Looks Like Inside an LLM and What We Can Learn From a Few Real Examples If you’ve ever used an LLM like ChatGPT or Claude, you’ve probably spent time thinking about how to write a good prompt. It might be as simple as asking a factual

Are LLMs Intelligent, or Just Playing the Assistant?

What if the real challenge in AI is not intelligence, but identity? Recent research suggests that language models may contain many possible personas, and that generating a response may begin with selecting a role. One of those roles appears to dominate modern systems. It is the familiar helpful assistant. In