This is according to research by the creator of ChatGPT, the bot that started the “AI”boom.

Is this what we want in datacenters sucking up our water?

If not, see a previous post for some bills in the Georgia legislature.

https://wwals.net/?p=69394

So-called AI hallucinates, no matter how good its training data –OpenAI 2025-09-18

Gyana Swain, Computerworld, September 18, 2025, OpenAI admits AI hallucinations are mathematically inevitable, not just engineering flaws,

In a landmark study, OpenAI researchers reveal that large language models will always produce plausible but false outputs, even with perfect data, due to fundamental statistical and computational limits.

OpenAI, the creator of ChatGPT, acknowledged in its own research that large language models will always produce hallucinations due to fundamental mathematical constraints that cannot be solved through better engineering, marking a significant admission from one of the AI industry’s leading companies.

The study, published on September 4 and led by OpenAI researchers Adam Tauman Kalai, Edwin Zhang, and Ofir Nachum alongside Georgia Tech’s Santosh S. Vempala, provided a comprehensive mathematical framework explaining why AI systems must generate plausible but false information even when trained on perfect data.

“Like students facing hard exam questions, large language models sometimes guess when uncertain, producing plausible yet incorrect statements instead of admitting uncertainty,” the researchers wrote in the paper. “Such ‘hallucinations’ persist even in state-of-the-art systems and undermine trust.”

The admission carried particular weight given OpenAI’s position as the creator of ChatGPT, which sparked the current AI boom and convinced millions of users and enterprises to adopt generative AI technology.

The rest of the article is well worth reading, as is the underlying paper.

Please note that the paper says, “However, a non-hallucinating model could be easily created, using a question-answer database and a calculator, which answers a fixed set of questions such as “What is the chemical symbol for gold?” and well-formed mathematical calculations such as “3 + 8”, and otherwise outputs IDK.” IDK is short for “I don’t know”.

So there are applications for which LLMs can be useful and even trustworthy.

But those are not the sorts of things for which most people are using ChatGPT and its cousins.

Some people advocate human reviewers to catch LLM hallucinations. But that’s not so easy, “These enterprise concerns aligned with broader academic findings. A Harvard Kennedy School research found that ‘downstream gatekeeping struggles to filter subtle hallucinations due to budget, volume, ambiguity, and context sensitivity concerns.’”

How can any budget be expected to cope with the breakneck rollout of LLMs?

How about we put a crimp in that rollout by denying some datacenters.

 -jsq, John S. Quarterman, Suwannee RIVERKEEPER®

You can help with clean, swimmable, fishable, drinkable, water in the 10,000-square-mile Suwannee River Basin in Florida and Georgia by becoming a WWALS member today!
https://wwals.net/donations/

Short Link: