![\"\"](\"https:\/\/world.edu\/wp-content\/uploads\/2025\/02\/Large-Language-Models.png\")
<\/p>\n<\/p>\n
AI has made it easier than ever to find information: Ask ChatGPT almost anything, and the system swiftly delivers an answer. But the large language models that power popular tools like OpenAI\u2019s ChatGPT or Anthropic\u2019s Claude were not designed to be accurate or factual. They\u00a0regularly \u201challucinate\u201d<\/a>\u00a0and offer up falsehoods as if they were hard facts.<\/p>\n Yet people are relying more and more on AI to answer their questions.\u00a0Half of all people<\/a>\u00a0in the U.S. between the ages of 14 and 22 now use AI to get information, according to a 2024 Harvard study. An analysis by\u00a0The Washington Post<\/a>\u00a0found that more than\u00a017% of prompts<\/a>\u00a0on ChatGPT are requests for information.<\/p>\n One way researchers are attempting to improve the information AI systems give is to have the systems indicate how confident they are in the accuracy of their answers. I\u2019m a\u00a0computer scientist<\/a>\u00a0who studies natural language processing and machine learning.\u00a0My lab<\/a>\u00a0at the University of Michigan has developed a new way of deriving confidence scores that improves the accuracy of AI chatbot answers. But confidence scores can only do so much.<\/p>\n Leading technology companies are increasingly integrating AI into search engines. Google now offers AI Overviews that appear as text summaries above the usual list of links in any search result. Other upstart search engines, such as\u00a0Perplexity<\/a>, are challenging traditional search engines with their own AI-generated summaries.<\/p>\n The convenience of these summaries has made these tools very popular. Why scour the contents of multiple websites when AI can provide the most pertinent information in a few seconds?<\/p>\n AI tools seem to offer a smoother, more expedient avenue to getting information. But they can also lead people astray or even expose them to harmful falsehoods. My lab\u00a0has found that<\/a>\u00a0even the most accurate AI models hallucinate in 25% of claims. This hallucination rate is concerning because other research suggests AI can\u00a0influence what people think<\/a>.Popular and problematic<\/strong><\/h4>\n
\n