True Positive Weekly #142
The most important artificial intelligence and machine learning news and articles
Hey, in this issue: one real reason AI isn’t delivering; how AI coding agents work; an introduction to speculative decoding; diffusion beats autoregressive; a tutorial on post-training GUI agents for computer use; robotic foundation models; three books to start or advance your career in AI; and more.
One real reason AI isn’t delivering: Meatbags in manglement
Google’s year in review: 8 areas with research breakthroughs in 2025
[Ars Technica] How AI coding agents work—and what to remember if you use them
[Nvidia] An introduction to speculative decoding for reducing latency in AI inference
[Apple] How even a single parameter can determine a LLM’s behavior
[CMU] Diffusion beats autoregressive in data-constrained settings
ChapterPal: Your personal AI tutor, helping you understand complex reading material
[Tutorial] Smol2Operator: Post-training GUI agents for computer use
[Video] Dexterous robotic foundation models
Enjoy the newsletter? Please, don’t forget to like or comment to help the newsletter grow.
Spot on. This issue is brilliant. That 'meatbags in manglement' point is exactly why AI isnt delivering, and it's so important to call it out. Thanks for this.