Research Explainers 7 min read

Long-Context Models Need Time to Consolidate

Language Models Need Sleep argues that long-context systems may require offline consolidation time, not merely larger memory, to reason over context they can no longer attend to. The analogy motivates architectures that compress experience between active reasoning cycles.
Digest 6 min read

Daily Digest - 2026-05-26

Cheaper code generation does not guarantee better products; durable gains depend on maintainable interfaces, parallel orchestration, measurable feedback loops, and attention systems that convert abundant output into coherent, useful progress.
Research Explainers 4 min read

AI Authorship Has Been Here Before

A legal history argues that AI authorship is not a new copyright problem but the latest form of mediated authorship. Earlier disputes over photography and mechanical creation reveal recurring legal tests for control and originality.
Research Explainers 7 min read

Train the Agent Where It Actually Runs

Polar shows how to train coding agents through the harnesses they already use instead of rebuilding those environments for reinforcement learning. The approach preserves realistic tools and feedback while making training operationally tractable.
Research Explainers 7 min read

Food AI Needs Knobs Beyond Recommendations

Epicure shows how ingredient embeddings can become navigable tools for cooking, menu design, and food AI. The system turns flavor relationships into controllable dimensions for substitution, exploration, menu planning, and creative composition rather than returning another opaque ranking.
Profile 8 min read

Lessons from Aidan Gomez

Aidan Gomez co-authored “Attention Is All You Need,” introducing the Transformer architecture, then founded Cohere to bring efficient, private language models to enterprises. His perspective tests architectural limits and prizes practical utility over pursuing artificial general intelligence.
Profile 7 min read

Lessons from Alan Kay

Alan Kay helped define personal computing through graphical interfaces and object-oriented programming at Xerox PARC, while his Dynabook anticipated laptops and tablets. His work asks whether system architecture and human learning have yet delivered computing’s real revolution.
Profile 9 min read

Lessons from Bill Buxton

Bill Buxton, Canadian computer scientist and designer, grounds human-computer interaction in low-fidelity sketching and the “Long Nose of Innovation,” showing why historical context, device ecosystems, and real human behavior matter when technologies slowly mature.

Choose your reading rhythm.

Start with a weekly briefing, add daily notes, or hear only when a durable essay or research update is ready.
You've successfully subscribed to Antoine Buteau
You've successfully subscribed to Antoine Buteau
Welcome back! You've successfully signed in.
Press ESC to close.