New Delhi: OpenAI’s latest language model, GPT-5, may be more powerful and accurate than its predecessors, but the company has warned users not to treat ChatGPT as their main source of information.
Nick Turley, Head of ChatGPT, said the AI chatbot should be used as a “second opinion” because it is still prone to mistakes, despite major improvements.
In an interview with The Verge, Turley admitted that GPT-5 continues to face the problem of hallucinations, where the system produces information that sounds believable but is factually wrong.
OpenAI says it has reduced such errors significantly, but the model still gives incorrect responses about 10 per cent of the time.
Turley stressed that achieving 100 per cent reliability is extremely difficult.
“Until we are provably more reliable than a human expert across all domains, we’ll continue to advise users to double-check the answers,” he said.
“I think people are going to continue to leverage ChatGPT as a second opinion, versus necessarily their primary source of fact,” he added.
Large language models like GPT-5 are trained to predict words based on patterns in huge datasets.
While this makes them excellent at generating natural responses, it also means they can provide false information on unfamiliar topics.
To address this, OpenAI has connected ChatGPT to search, allowing users to verify results with external sources.
Turley expressed confidence that hallucinations will eventually be solved but cautioned that it will not happen in the near future.
“I’m confident we’ll eventually solve hallucinations, and I’m confident we’re not going to do it in the next quarter,” he said.
Meanwhile, OpenAI continues to expand its ambitions. Reports suggest the company is developing its own browser, and CEO Sam Altman has even hinted that OpenAI could consider buying Google Chrome if it were ever put up for sale.
(IANS)
