CAN 10,000 AI PAPERS PREDICT WHICH TOPICS WILL GROW SIX MONTHS LATER?

Archived study

Tracks the shares of five AI research themes across three supplied arXiv datasets. It compares changes before and after July 2024 using percentages and uncertainty rather than raw counts alone.

Interpretation and limits

The findings are limited to the supplied files, theme definitions, and comparison periods. Multiple comparisons and uncertain smaller changes prevent treating every apparent increase as a reliable prediction.

Package and provenance

The original .science export and matching poster are preserved byte-for-byte from the upstream case. The package contains the conversation, session metadata, and the resources inventoried in its internal manifest. This introduction summarizes the archived record; the research, external sources, and model evaluations were not rerun for this import.