More and more people are turning to chatbots to search the web. Like the search engines of old, they rank information to determine what’s most relevant.
But researchers are finding these systems can be gamed.
One example that caught our eye involved a researcher who created a fake deodorant brand — and got ChatGPT to recommend it over real products.
We wanted to understand more about how AI chatbot search results can be manipulated, so we called up someone who studies these things for a living: Xiyang Hu, an assistant professor at Arizona State University, where he leads the Generative Learning and Augmented Decision Lab.
“Someone Hid It!'': Query-Agnostic Black-Box Attacks on LLM-Based Retrieval — a study published for the International Conference on Machine Learning
Reddit comments are 'foundational' to training AI models, COO says — from Business Insider
Language Models Often Favor Their Own Text, Revealing a New Bias in AI — NYU Center for Data Science
Apple
Spotify
Amazon
RSS