ChatGPT Likes to Make Up Its

Own Stories, Its Creator Confused

By Many People’s Trust

Jakarta, CNBC Indonesia –The development of AI has led many people to place greater trust in the technology, even leading them to use it without double-checking its accuracy.

However, OpenAI CEO Sam Altman cautioned against placing complete trust in ChatGPT. He admitted the popular chatbot is prone to hallucinations and is unreliable.

“People have a high level of trust in ChatGPT, which is interesting, because AI hallucinates. This should be a technology you don’t trust too much,” Altman said in a podcast quoted by Mint, Tuesday (July 8, 2025).

Hallucinations in AI chatbots refer to the product making up stories with extreme confidence. This occurs for several reasons.

For example, biased chatbot training data, lack of real-world knowledge, pressure to consistently produce responses, and creating predictive text.

Altman also reiterated that his children are not smarter than AI. However, he assured them that they will surpass their parents’ achievements today.

“But they’ll grow to be much more capable than us and be able to do things we can’t even imagine,” he said.

Altman was also asked about the possibility of advertising being included in ChatGPT. He said it would be difficult to implement.

However, Altman said he wasn’t opposed to the idea. The 40-year-old even praised the ads displayed on Instagram.

“I think Instagram ads are pretty cool; I buy a lot of stuff from there. But I think it would be hard to do it right,” Altman explained.

He explained that ads could be implemented without completely disrupting the user experience. However, the burden of proof is very difficult.

“The burden of proof would have to be very high, and it would have to be very beneficial to the user, and it certainly wouldn’t compromise the LLM,” he said.

SOURCE : CNBC INDONESIA