AI can answer almost any question you ask, but there is also a lot of nonsense involved. Now ChatGPT can finally admit that it has no idea about a topic. And this is an important change for the use of artificial intelligence.
Large language models (LLMs) have had a long discussion regarding telling the truth. Because anyone who has worked with artificial intelligence knows that AI tends to hallucinate: at some point, the chatbot invents facts, data, or even quotes.
Now the English-language magazine Techradar reports that Kol Tregaskes posted a significant change in conversation with AI on X.com. Because ChatGPT 5.0 admits on its own that it has no idea or doesn’t know an answer.
And this is a capability that children develop at the age of 4-5 years, reports Kindergesundheit-info.de.
Instead of making things up, AI can admit that it has no information
What is the problem with inventions? Hallucinations have been a problem for AI chatbots since the technology was introduced a few years ago. However, the big problem is different: AI displays the same level of confidence in its answers regardless of their accuracy. It offers you false information and invented quotes, even if they may be completely wrong or fabricated. However, these are not labeled as invented; AI presents you with these faulty pieces of information as correct content.
The reasons for such hallucinations can be diverse, explains the Fraunhofer Institute: Faulty data sources or a training method can quickly lead to an AI spouting nonsense.
Why is the new response so important? An open admission of ignorance replaces fictional filler words. At first glance, it may be disappointing that AI cannot provide a perfect or correct answer. However, this admission makes AI appear much more trustworthy and reduces the chance of fabricated answers.
Trust is important, as many users frequently question the answers provided by artificial intelligence or have already fallen for false information.
An AI tool erases a developer’s work in seconds. The AI ignored all backups and acted unilaterally. The AI itself explained that it was a catastrophic error and that the data could not be recovered: An AI ignores all measures and unilaterally deletes a developer’s entire work in seconds
