If you have spent any real time with Large Language Model (LLM) AI systems like Grok or ChatGPT, you probably know the pattern:
‘OMG this is AMAZING!’
‘Okay, not ‘amazing’ but still really useful!’
‘Can come in handy sometimes.’
‘I probably should double-check all of this.’
‘Really?’
Listen here on Soundcloud:
Yes, well … A LOT of people in here, people with information technology systems experience, engineering experience and programming experience — Have been trying hard to tell you that the “I” part of “AI” is a myth at best and an outright lie more likely. But hey …
Don’t listen to us old pharts, stick your hand on a hot stove and find out for yourself. It seems that’s the only way people learn.
A colleague of mine referred to AI capabilities to be likened to that of an over-eager grad student. Unfortunately, such individuals spew lots of garbage and useless information. It is the job of the “professor” to winnow the chaff from the wheat. These LLMs are obviously not being winnowed properly or sufficiently.
When I was a fledgling student learning the basics of programming on a Commodore 64, the GIGO acronym was introduced: Garbage In, Garbage Out. That definition still applies to what a so-called AI, which is just the latest sexy headline for LLM (Large Language Model), can process. They are nothing more than highly-complex parameterized functions that optimize probable responses to input. That input starts with the data that is used to create the billions of parameters in the function, and when people provide it garbage input data, we get garbage output responses. These are not thinking machines. Regardless of what… Read more »
I’ll borrow. From scarecrow “Some people without brains do an awful lot of talking.”
Chat doesn’t understand anything.. It’s incapable of understanding anything. It has glosolalia in pixels.
Nice word. Thank you for its introduction to my vocabulary. It succinctly describes the fundamental problem of the current, over-publicized term: artificial intelligence.
They do make money; in ‘investment’. It’s a Pyramid scheme, of sorts.
Be careful guys! We don’t want ChatGPT to be called a ‘radical’ because it deactivated some carbon-based units.
I only use AI when I am searching for an answer. With using Duck Duck Go, their AI is at the top and summarizes the answers to help out. Although I also saw two buttons: “News on the left” and “News on the right” when I searched about the (false) story of Trump sending ICE to the Olympics (not directly, but DHS is helping the State Dept. with security for our athletes, and ICE is part of DHS–so no, he did not say “I’m sending ICE to Italy.”). I thought those buttons were interesting but did not use them. I… Read more »
Thanks for pointing out the DDGo alternate search feature under the “Googles” button. Interesting that “News from right” was updated in 2023, while “News from left” was updated 2025.
You are welcome. And thank you for taking one for the team and checking out the “News from the right” and “News from the left” buttons.
One thing I never hear mentioned about AI that is most definitely a serious drawback and potential problem, is that at this stage in its development it has all of recorded human history to draw on scan, collate and regurgitate in order to satisfy your requests. However, a major portion of those histories are wrong as recorded and AI cannot know that! Going forward it only gets worse as the information available for AI to draw from becomes an ever-higher percentage of its own “interpretations” of what it has spewed forth before, thus at some point in the future, not… Read more »