causal

Models Lack Memory Scaffolding for Salient Facts

Current models are not good at knowing which memories to store because their training only optimizes next-word prediction; unlike humans who actively construct memorable summaries, models lack the opportunity to do so, and memory scaffolding remains very primitive.

causalpending

Speaker

Trenton Bricken

Evidence Quote

“Memory scaffolding in general is just very primitive right now.”

Source

AMA: career advice given AGI, how I research ft. Sholto & Trenton— Dwarkesh Patel Podcast
Created: 6/13/2026, 3:36:46 AM

Your notes

Loading notes...