AI research tools risk entrenching academic biases through the Matthew effect

Summary

This article examines how AI-powered academic research tools (like Google's Deep Research, Anthropic's Research, and OpenAI's deep research) are trained on scholarly literature that already suffers from unequal attention dynamics, particularly the Matthew effect (where highly-cited works attract even more citations). The author, Jefferson Pooley, argues that these AI tools will inadvertently reproduce and amplify existing biases in academic hierarchies, making them harder to detect and audit. The core concern is that AI summarization and ranking systems will entrench the advantage of already-prominent scholars and works, while further marginalizing less-cited but potentially valuable research.