Aldous Huxley's Deep Reflection(thereader.mitpress.mit.edu)

90 points by torritest 3 years ago | 51 comments

How Typing Transformed Nietzsche's Consciousness(thereader.mitpress.mit.edu)

61 points by bookofjoe 1 year ago | 25 comments

The Many Sides of Erik Satie(thereader.mitpress.mit.edu)

153 points by anarbadalov 1 year ago | 36 comments

The hallucinatory thoughts of the dying mind(thereader.mitpress.mit.edu)

142 points by anarbadalov 1 year ago | 93 comments

J.G. Ballard: My Favorite Books(thereader.mitpress.mit.edu)

149 points by anarbadalov 2 years ago | 42 comments

A Man Who Saw Humanity from Two Billion Years Away(thereader.mitpress.mit.edu)

61 points by samizdis 2 months ago | 14 comments

Claude’s memory architecture is the opposite of ChatGPT’s(shloked.com)

408 points by shloked 1 year ago | 216 comments

Blog post analyzing how Claude's memory system works, retrieving from raw conversation history on demand, and contrasting it with ChatGPT's memory design.

AI has access to a vastly larger working memory than the human brain(davidepiffer.com)

524 points by rzk 2 months ago | 452 comments

A Substack essay by Davide Piffer arguing that AI's edge over mathematicians comes from a vast symbolic working memory rather than superior reasoning.

Aging brains blend memories together instead of just forgetting them(studyfinds.com)

323 points by mdp2021 1 month ago | 129 comments

StudyFinds article on Binghamton University research suggesting aging brains blend similar memories together rather than simply forgetting them.

Human brains are preconfigured with instructions for understanding the world(news.ucsc.edu)

414 points by XzetaU8 11 months ago | 284 comments

UC Santa Cruz news article reporting research suggesting early developing human brains are preconfigured with instructions for understanding the world.

LLMs use a surprisingly simple mechanism to retrieve some stored knowledge(news.mit.edu)

336 points by CharlesW 2 years ago | 121 comments

MIT News article on research showing LLMs often retrieve stored facts via simple linear functions, and a technique to probe what a model knows about subjects.

Neural networks are clean(already-clean.goodfire.com)

768 points by themgt 11 days ago | 300 comments

An informal research essay by Nick Cammarata at Goodfire arguing polysemanticity stems from fixable issues, not superposition, making neurons interpretable at scale.