Catastrophic forgetting

Oct 08, 2026

$450K TO TRAIN A FRONTIER AI MODEL?

Thomson Reuters CTO Joel Hron explains why the company built Thomson, a domain-specific frontier model adapted from open-weight checkpoints like Qwen 3.5, using decades of proprietary legal and tax content and expert judgment. The conversation covers the training lifecycle from realignment and continual pre-training to agentic RL, the economics ($450K final run versus roughly $40M total), team structure, CoCounsel orchestration, verification challenges in legal and tax, and the future balance between open-weight and frontier models.

Aug 18, 2026

Rich Sutton and Khurram Javed: Why AI Models Stop Learning, and How to Start It Again

Rich Sutton and Khurram Javed discuss their radical vision for AI at Oak Lab, advocating for truly continual learning agents that learn from their own experience, rejecting synthetic data due to the "Big World Hypothesis," and outlining a path to overcome catastrophic forgetting with "continual backprop" for a trillion-parameter, self-maintaining mind, while critiquing LLMs as only a fraction of intelligence.

Jul 23, 2026

Coding Agents Are Secretly General Agents

Jay Hack, head of AI at ClickUp, discusses the evolution of AI from early computer vision to generalist coding agents, highlighting how 'positive transfer' makes coding an 'AGI-complete' domain. He delves into the brutal economics of AI startups facing foundation model giants, ClickUp's strategy for convergence and first-party data as a moat, and the challenges of verifiability and catastrophic forgetting. The conversation also explores LLMs at the scientific frontier, the 'car wash test' revealing limits of world models, and speculative future applications like LLM resorts and game integration.

Dec 29, 2025

Memory in LLMs: Weights and Activations - Jack Morris, Cornell

This talk explores the limitations of current methods for providing knowledge to LLMs, such as large context windows and Retrieval-Augmented Generation (RAG). The speaker argues that the future lies in training knowledge directly into the model's weights. This is achieved through a combination of generating large synthetic datasets from small amounts of source material and using parameter-efficient fine-tuning (PEFT) techniques like LoRA to avoid catastrophic forgetting. The goal is to create more capable, personalized, and efficient models by fundamentally altering how they store and access information.