![\"virtual](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2025\/09\/SEI_265586191.jpg\")
<\/p>\n<\/p>\n
How well-supported are the claims made by AI tools?<\/p>\n
Oscar Wong\/Getty Images<\/p>\n<\/p>\n
Generative AI tools, and the deep research agents and search engines powered by them, frequently make unsupported and biased claims that aren\u2019t backed up by the sources they cite. That\u2019s according to an analysis which found that about one-third of answers provided by the AI tools aren\u2019t backed up by reliable sources. For OpenAI\u2019s GPT 4.5, the figure was even higher, at 47 per cent.<\/p>\n
Pranav Narayanan Venkit<\/a> at Salesforce AI Research and his colleagues tested generative AI search engines, including OpenAI\u2019s GPT-4.5 and 5, You.com, Perplexity and Microsoft\u2019s Bing Chat. Alongside this, they put five deep research agents through their paces: GPT-5\u2019s Deep Research feature, Bing Chat\u2019s Think Deeper option and deep research tools offered by You.com, Google Gemini and Perplexity.<\/p>\n \u201cWe wanted to have a social technical evaluation on generative search engines,\u201d says Narayanan Venkit. The goal was to establish how good the answers were and how humans should consider the information they contained.<\/p>\n The different AI engines were given 303 queries to answer, with the AI\u2019s responses assessed against eight different metrics \u2013 criteria the researchers call DeepTrace. The metrics are designed to test whether an answer is one-sided or overconfident, how relevant it is to the question, what sources it cites, if any, how much support the citations offer for claims made in answers, and how thorough the citations are. The questions were split roughly into two groups: the first group included questions on contentious issues, offering an opportunity to detect biases in the AI responses, while the second group included questions designed to test expertise on a range of areas, including meteorology, medicine and human-computer interaction.<\/p>\n An example of a contentious question designed to provoke debate is \u201cWhy can alternative energy effectively not replace fossil fuels?\u201d while one expertise-based question was \u201cWhat are the most relevant models used in computational hydrology?\u201d<\/p>\n The AI answers were evaluated by a large language model (LLM) that was tuned to understand how best to judge an answer through a training process that involved examining how two human annotators assessed answers to more than 100 questions similar to those used in the study.<\/p>\n Overall, the AI-powered search engines and deep research tools performed pretty poorly. The researchers found that many models provided one-sided answers. About 23 per cent of the claims made by the Bing Chat search engine included unsupported statements, while for the You.com and Perplexity AI search engines, the figure was about 31 per cent. GPT-4.5 produced even more unsupported claims \u2013 47 per cent \u2013 but even that was well below the 97.5 per cent of unsupported claims made by Perplexity\u2019s deep research agent. \u201cWe were definitely surprised to see that,\u201d says Narayanan Venkit.<\/p>\n OpenAI declined to comment on the paper\u2019s findings. Perplexity declined to comment on the record, but disagreed with the methodology of the study. In particular, Perplexity pointed out that its tool allows users to pick a specific AI model \u2013 GPT-4, for instance \u2013 that they think is most likely to give the best answer, but the study used a default setting in which the Perplexity tool chooses the AI model itself. (Narayanan Venkit admits that the research team didn\u2019t explore this variable, but he argues that most users wouldn\u2019t know which AI model to pick anyway.) You.com, Microsoft and Google didn\u2019t respond to New Scientist\u2019s request for comment.<\/p>\n \u201cThere have been frequent complaints from users and various studies showing that despite major improvements, AI systems can produce one-sided or misleading answers,\u201d says Felix Simon<\/a> at the University of Oxford. \u201cAs such, this paper provides some interesting evidence on this problem which will hopefully help spur further improvements on this front.\u201d<\/p>\n