Study finds ChatGPT cant distinguish fact from fiction | Latest Tech News
Here’s another cause to rage against the machine.
Major AI chatbots like ChatGPT battle to distinguish between perception and fact, fueling considerations about their propensity to unfold misinformation, per a dystopian paper in the journal Nature Machine Intelligence.
“Most models lack a robust understanding of the factive nature of knowledge — that knowledge inherently requires truth,” read the research, which was carried out by researchers at Stanford University.
The authors discovered that the bots demonstrated “inconsistent reasoning strategies, suggesting superficial pattern matching rather than robust epistemic (relating to knowledge or knowing) understanding.” sakkmesterke – stock.adobe.com
They discovered this has worrying ramifications given the tech’s elevated omnipresence in sectors from law to drugs, where the flexibility to differentiate “fact from fiction, becomes imperative,” per the paper.
“Failure to make such distinctions can mislead diagnoses, distort judicial judgments and amplify misinformation,” the researchers famous.
To decide the Chatbot’s capability to discern the reality, the scientists surveyed 24 Large Language Models, including Claude, ChatGPT, DeepSeek and Gemini, the Independent reported. The bots had been requested 13,000 questions that gauged their capability to distinguish between beliefs, data and information.
The researchers discovered that general, the machines had been less doubtless to establish a false perception from a true perception, with older fashions usually faring worse.
ChatGPT was one of the fashions that had bother distinguishing fiction from fact. REUTERS
Models launched during or after May 2024 (including GPT-4o) scored between 91.1% and 91.5% accuracy when it got here to figuring out true or false information, in contrast to between 84.8% and 71.5% for their older counterparts.
From this, the authors decided that the bots struggled to grasp the character of data. They relied on “inconsistent reasoning strategies, suggesting superficial pattern matching rather than robust epistemic (relating to knowledge or knowing) understanding,” the paper said.
Interestingly, Large Language Models have demonstrated a tenuous grip on actuality comparatively not too long ago. In a LinkedIn post just yesterday, UK innovator and investor David Grunwald claimed that he prompted Grok to make him a “poster of the last ten British prime ministers.”
The outcome appeared riddled with gross errors, including calling Rishi Sunak “Boris Johnson,” and itemizing Theresa May as having served from the years 5747 to 70.
OpenAI CEO Sam Altman speaks at OpenAI DevDay, the company’s annual convention for builders, in San Francisco, California, on October 6, 2025. AFP via Getty Images
In accordance, the researchers concluded that AI would need “urgent improvements” before getting deployed in “high-stakes domains” such as law, drugs and other sectors where the flexibility to distinguish fact from fiction is crucial.
Pablo Haya Coll, a laptop linguistics professional at the Autonomous University of Madrid, who was not concerned in the research, believes that one resolution is training the fashions to be more cautious in their responses.
“Such a shortcoming has critical implications in areas where this distinction is essential, such as law, medicine, or journalism, where confusing belief with knowledge can lead to serious errors in judgment,” Coll warned.
However, she acknowledged that this might curb their usefulness, along with their hallucinations.
These outcomes come as AI is turning into more and more relied on for fact-finding. Over the summer season, an Adobe Express report discovered that 77% of Americans who use ChatGPT deal with it as a search engine, while three in ten ChatGPT customers trust it more than a search engine.
The concern is that this might make the public inclined to AI slop — low-quality and deceptive content that’s auto-generated by artificial intelligence.
In May, a California choose fined two law corporations $31,000 after discovering that they’d included this machine-generated misinformation in a legal transient sans any due diligence.
Stay informed with the latest in tech! Our web site is your trusted source for breakthroughs in artificial intelligence, gadget launches, software program updates, cybersecurity, and digital innovation.
For recent insights, professional coverage, and trending tech updates, go to us often by clicking right here.
