Claude's Cycles [pdf]

www-cs-faculty.stanford.edu

March 3, 2026

1 min read

🔥🔥🔥🔥🔥

72/100

Summary

Posted by fs123. Score: 248 points. Comments: 122.

Coverage & discussions

Hacker News · www-cs-faculty.stanford.edu

Historical count: 5 · 5 saved in sample · Historical sample

Analysis ready

Community Sentiment

Positive

Positives

The ability of LLMs to scale probability distributions allows for rapid problem-solving, making advanced techniques accessible to a broader audience and enhancing expert productivity.
There's potential for AI to tackle complex problems like connecting general relativity and quantum physics, which could lead to groundbreaking discoveries in science.
LLMs possess superhuman knowledge, enabling them to make unique connections and engage in tireless trial and error, which could yield innovative solutions previously unseen by humans.

Concerns

The fundamental question remains whether LLMs truly 'think' or simply predict the next word based on probability, raising concerns about the depth of their problem-solving capabilities.