← Browse

Session brief

Jack Morris: Stuffing Context is not Memory, Updating Weights is

63 min

Overview

This talk challenges the conventional approaches of stuffing context or using Retrieval Augmented Generation (RAG) for AI knowledge integration. The speaker argues that these methods are fundamentally limited by cost, speed, and reasoning capabilities. Instead, the focus shifts to training information directly into model weights, proposing this as a more effective and scalable solution for building AI systems that truly know and can reason with specific data.

Who should watch

Key takeaways

Notable quotes

*Stuffing context is not memory, updating weights is.*
*The models we use are transformers. The real problem with transformers comes in this one little box right here called self-attention.*
*Embeddings are the file system of today and they're not the file system of the future.*

Watch on YouTube →

Unofficial community note. Prefer the recording for nuance.