Predicting the Past: Conversing with Antiquity

It shows how narrow, high-accuracy research models can be exposed to non-programmers through an agent skill layer, turning restoration, dating and geolocation of damaged Greek and Latin inscriptions into a conversational workflow. For anyone building domain agents, it is a working reference for wrapping specialist models instead of defaulting to a general-purpose LLM.

• agent skill — A reusable instruction file that teaches an agent how to do one job well — the procedure, the tools, and what counts as done.
• LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.

• postDeepMind's AlphaGenome Atlas already finding disease-causing DNA variants
• postGoogle DeepMind Launches Gemini 3.8 Live and Extended Thinking Models
• postWeatherNext 3 moves to hourly forecasts at 5km resolution
• article🔵 It shows strong gains over 3.6 Flash in key coding tasks like debugging and i

• articleThree Years from GPT-3 to Gemini 3Ethan Mollick
• articleAn Opinionated Guide to Which AI to Use to Do Stuff (Summer 2026 Edition)Ethan Mollick
• articleThe OverhangEthan Mollick

Checking sign-in…

Loading comments…