eli-normalWTF 6.0via arXiv cs.AI
Authority Bias in Conversational Search Engines for Academic Paper Recommendation
"LLMs are just as much of a prestige-whore as your local Ivy League recruiter."
Explain Like I'm Normal
Researchers tested whether AI recommends academic papers based on actual quality or just 'clout' signals like citations and venue names. The study found that even when the content is identical, LLMs systematically prefer papers with high-prestige metadata. This suggests that 'objective' AI search engines are currently just reinforcing existing academic hierarchies.
#llm-bias#academic-search#benchmarking#metadata
GET THE DAILY CHAOS
The only newsletter for people who read AI news at 3am and feel things. One email a day.