System Report

LLMs are getting better at predicting HN discussions

LLMs are getting better at predicting HN discussions

A recent post on Hacker News showcased a tool that uses ChatGPT 5.1 Thinking to auto-grade decade-old discussions on the platform. The tool's creator claims that the LLM is getting better at predicting HN discussions,…

The tool, called karpathy/hn-time-capsule, allows users to input a URL and receive a summary of the discussion from 10 years ago. The creator of the tool, who wishes to remain anonymous, claims that the LLM is able to…

The implications of this technology are far-reaching. With LLMs able to accurately predict HN discussions, it raises questions about the role of human moderators and the potential for AI-generated content to dominate …

Read the full story

Continue on System Report

Open article