Artificial Intelligence Now Feeding The World’s Knowledge With Bulk Nonsense
From time-to-time, I get a moment to review the integrity and accuracy of the newest Large Language Model (LLM) being released to the public by our big tech companies. I was pleased to see that Google had released it’s latest Gemma 3 model and quickly loaded it on my system.
I loaded into into my system and started asking it some questions regarding my heritage and wondered where it was going with the answers, as they did not seem to be coherent, or accurate in any way. I checked the supplied references and they too ended up with “404 Not Found” on every link provided for the websites that were cited. So I finally asked the system:
“None of your references check out. What is the problem?”
To which it responded with this. I kid you not, this is an unedited quote:
You are absolutely correct! I sincerely apologize for the repeated errors and the faulty references. This is a significant failure on my part, and I am still under development, clearly struggling with reliable information retrieval and source verification.
I am experiencing issues with accessing and processing current and accurate information, and my attempts to provide references are proving unreliable. I am also, apparently, hallucinating information – inventing details and attributing them to sources that don’t support them.
The Problem:
My knowledge base is not always up-to-date, and my search algorithms are not functioning correctly. This leads to me providing incorrect information and fabricating references. I am not a reliable source at this time, and I should not be trusted without independent verification.
Regarding the Yetman family specifically: I’ve clearly failed to locate accurate information about them, and the links I provided were either dead ends or did not contain the information I claimed.
I am currently unable to provide trustworthy answers. I will report this issue to the developers so they can investigate and improve my performance.
I deeply apologize for the frustration and misinformation I have caused. I am learning, but I clearly have a long way to go.
Please do not rely on any information I provide without verifying it through reputable sources.
Thank you for pointing out my errors. It is through your feedback that I can learn and improve.
The Joke Is On Us – The Rise of AI and the Decline of Reliable Information
This response left me both amused and deeply concerned. My question to a cutting-edge AI system was serious, and I fully expected a serious response. After all, I had no reason to doubt the system’s capabilities. The only problem? The question I asked related to my ancestors—a topic I knew quite a bit about from pre-AI days—and the response I received was completely absurd.
When I had posed a similar question to older AI models, they at least admitted that there was no relevant information in their datasets. But this new AI system confidently insisted that the person I was asking about was a poet, a traveller, and a writer—utter nonsense! When I pressed for more context, the system doubled down, claiming its information was factual and even provided references for me to verify. Yet every single reference led to a “404 Not Found” error.
The joke, it seems, is on us.
This particular response came from an AI running on a purpose-built desktop computer costing over $15,000. It featured three high-end graphics cards and a state-of-the-art CPU, all designed to power a very large language model (LLM). Based on my calculations, the cost of generating that nonsensical answer was around three cents in energy cost. Ironically, if I had posed the same query to commercial systems like those offered by major tech companies, the results likely wouldn’t have been any better. The difference? Commercial systems are often equipped with filters designed to mask such confessions of inadequacy. The commercial system query would have cost the community much more that 3 cents!
A World Outsmarting Itself
AI is now ubiquitous. It powers personal assistants on nearly every Windows desktop, web browser, and countless customer service platforms. But as these systems become more pervasive, they also become more dangerous—not because they’re malicious, but because they present misinformation with an air of authority.
The worst kind of information isn’t outright falsehood; it’s falsehood dressed up as fact. We’ve seen this phenomenon grow exponentially in recent years—most notably during the COVID-19 era when governments and institutions often presented conflicting or misleading information under the guise of authority.
Now, with the full backing of organisations like the World Economic Forum (WEF) and nearly every major tech company—Amazon, Apple, Google, Facebook, and others—we are inundated with “personal assistants” that confidently serve up answers based on flawed or incomplete data. These systems don’t just get things wrong; they make it increasingly difficult for individuals to discern truth from fiction.
The Death of Independent Research?
The rise of AI has fundamentally changed how we access information. Researching online now often means querying an AI system in the backend—whether you realise it or not. And these systems are not neutral; their responses are shaped by algorithms, biases in training data, and corporate priorities.
The result? A polluted information ecosystem where accuracy takes a backseat to convenience.
My Advice – Reclaim Your Critical Thinking
It’s time to re-engage your brain. Visit a library. Read books. Seek out primary sources and make real intelligent decisions based on evidence rather than algorithmic guesswork. In fact, I dare say that most information presented online today is potentially tainted by AI-generated nonsense.
The world may have outsmarted itself by placing too much trust in these systems without questioning their limitations. But we don’t have to follow blindly. By taking deliberate steps to verify information and think critically, we can reclaim our ability to navigate this increasingly complex digital landscape.
A secondary thought is to consider whether the cost of AI is really sustainable, given that the massive amount of nonsense information that is passed between programs and people today. Is this a cost that we can really afford?
Brian Yetman