Page 8 of 14 · Back to latest writing
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.
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.
SkillOpt treats reusable agent skills as operating assets that can be trained, validated, and reused. It shows how procedural instructions can be optimized against tasks, checked for regressions, and distributed with evidence.
Constraint decay explains why AI coding agents struggle when backend work carries real architectural and database rules. The paper traces how requirements fade across long trajectories and proposes stronger constraints, memory, and verification.
Yann LeCun is a Turing Award-winning computer scientist and architect of convolutional neural networks whose AI vision prioritizes world models, self-supervised learning, and open-source intelligence while questioning the industry’s reliance on scaling large language models and AGI hype.
Geoffrey Hinton is a pioneering neural-network researcher whose five-decade journey runs from early advocacy of connectionism to urgent warnings about existential risk, tracing a profound change in how he understands intelligence, computation, and humanity’s technological future.
Scott Stevenson, co-founder and CEO of Spellbook and a computer engineer by training, brings lean experimentation to legal technology, exploring how large language models can improve contract drafting through a philosophy of human-AI augmentation.
Dean W. Ball, a technology policy fellow at the Mercatus Center at George Mason University, examines AI, governance, and reindustrialization, advocating state support for technological diffusion and infrastructure instead of restrictive regulation aimed at individual models.