UC Riverside Study: LLMs Prioritize Logic Over Human Qualities
TL;DR. A collaborative study by UC Riverside researchers determined Large Language Models favor logic and factual consistency when responding to subjective questions online. - LLM responses contrast with human-written web pages, which incorporate emotion, experience, and ethics. - Computer scientists presented these findings at WebSci26, comparing ChatGPT and Gemini against Google and Bing. - Researchers warn that internet content may lose its human nature if AI-assisted search continues to dominate.
- UC Riverside study finds LLMs like ChatGPT and Gemini prioritize logic in subjective responses.
- Human-written web content uses a richer mix of emotion, lived experience, and ethics.
- The analysis compared AI system responses with traditional web search results to controversial questions.
- Researchers suggest overwhelming reliance on AI for information could diminish the internet's human reflection.