Researcher at University of Zurich
Swiss researcher who criticizes the paper's use of LLM-based annotation and statistical methodology
How media typically covers Aleksandra Urman
Directly quoted in these articles
Research by Salesforce AI Research found that approximately one-third of answers from generative AI search engines lack reliable source support, with OpenAI's GPT-4.5 reaching 47% unsupported claims and Perplexity's deep research agent at 97.5%.
“Swiss researcher who criticizes the paper's use of LLM-based annotation and statistical methodology”