OpenAI Chief Executive Officer Sam Altman has issued a cautionary statement on the unreliability of AI tools like ChatGPT, emphasising that they can construct plausible but incorrect outputs—a phenomenon known as “hallucination.” Speaking on the debut episode of OpenAI’s official podcast, Altman expressed concern over the public’s current perception of artificial intelligence as an almost infallible source of truth.

Altman remarked that he finds it “interesting” that people put a “very high degree of trust” in ChatGPT, remarking further that “AI hallucinates. It should be the tech that you don’t trust that much.” He highlighted the inherent tension between AI’s conversational fluency and its susceptibility to generating misleading information, despite its polished delivery.

In a candid personal anecdote, Altman explained how he relied heavily on ChatGPT during the early stages of parenthood. He shared that he frequently used the platform to seek guidance on matters such as diaper rash care and infant nap routines. While he acknowledged its utility, he admitted, “I had to remind myself it doesn’t always get it right,” underscoring the limits of AI in offering dependable advice.

Altman’s remarks come at a time when generative AI systems are increasingly woven into everyday life, being used for research, homework, legal summaries, and even patient diagnosis. He cautioned that blind reliance on these systems presents significant risks. “It’s not super reliable,” he said, warning users to approach AI with critical oversight, especially in contexts with high stakes.

He further warned that while AI is a powerful assistant, it must remain under human supervision. This plea for caution gains urgency against a backdrop of growing legal and ethical scrutiny. OpenAI is currently facing lawsuits alleging unauthorised use of copyrighted content to train its models.

Altman also touched on emerging AI trends, cautioning that as AI becomes more deeply embedded in society, new challenges will arise. These include concerns around privacy, potential parasocial relationships with AI agents, and the need for robust guardrails to regulate use. “Again, I suspect this is not all going to be good… society will have to figure out new guardrails,” he noted.

Experts in the AI ethics space confirm the potential dangers of over-reliance. The term “hallucination” highlights how language models can fabricate detailed but false information in a persuasive tone, making it difficult for users to discern fact from fiction without verification.

Altman’s transparent reflection underscores a crucial paradox: while AI offers unparalleled convenience and efficiency, its outputs remain fallible. The CEO’s message is clear—use AI as a tool, not a definitive authority. As generative AI continues to evolve and integrate into diverse sectors, from academia to healthcare, public awareness and critical scrutiny are vital.

Be a part of Elets Collaborative Initiatives. Join Us for Upcoming Events and explore business opportunities. Like us on Facebook , connect with us on LinkedIn and follow us on Twitter.

"Exciting news! Elets technomedia is now on WhatsApp Channels Subscribe today by clicking the link and stay updated with the latest insights!" Click here!