Blog
Building with AI in practice: language models, agents, prompting, and what the research says about using them well.
Underneath the chat window, a language model does one thing: guess the next token. How that single trick, scaled far enough, turns into something you can talk to, and the honest argument over whether any of it counts as understanding.
How to catch confident-but-wrong AI output before it bites you. A short, practical checklist plus the reason hallucinations happen in the first place.
Concrete ways an AI agent like Claude fits into a research workflow today: auditable literature work, pipelines with built-in human stop-gates, and where agents still fall short.
Vibe research means you set the direction and AI agents do the execution, while the judgement and accountability stay yours. What it is, why I'm optimistic, and the failure mode to watch for.
Modern AI looks like it arrived in November 2022. It's really the tail end of a ninety-year chain of ideas, almost none of them built by people aiming at anything like it. A walk from Turing to ChatGPT.
Instructing AI has moved from wording a prompt, to curating context, to engineering the loop the agent runs in. Some of these are real terms; some are barely a year old.
Vibe coding changes who gets to build software. Here's what it actually is.