Scott Alexander discusses recent breakthroughs in AI interpretability, explaining how researchers are beginning to understand the internal workings of neural networks. Longer summary

Scott Alexander explores recent advancements in AI interpretability, focusing on Anthropic's 'Towards Monosemanticity' paper. He explains how AI neural networks function, introduces the concept of superposition where fewer neurons represent multiple concepts, and describes how researchers have managed to interpret AI's internal workings by projecting real neurons into simulated neurons. The post discusses the implications of this research for understanding both artificial and biological neural systems, as well as its potential impact on AI safety and alignment. Shorter summary

A collection of links covering topics from drug regulation and genetic testing to tax reform and AI developments, with Scott's commentary on each item. Longer summary

This is a wide-ranging links post covering various news items and studies. Topics include FDA regulations and drug pricing, genetic testing advances, Russian Facebook ads from the 2016 election, studies on education and intelligence, tax reform effects on graduate students, and various scientific developments. The post moves quickly between topics, providing brief commentary and often linking to additional sources. Several items focus on medical and healthcare issues, from insulin prices to studies on pain medication effectiveness. The post maintains Scott's characteristic mix of intellectual curiosity and light humor throughout. Shorter summary