YooLooKool

I want to learn more about artificial intelligence. I’m a naturally curious person and I’m very, very curious about this new application or technology.

In so doing, I’ve been reading about it, asking questions about it, and conducting dialogues with a number of the AI programs. I use GROK, chatGPT, Claude and Gemini. And recently I’ve signed up for… Plurilog.

And attached is the Plurilog discussion after. It is at the bottom of this epost.

This is a fairly new app and I do LOVE how the different AIs are playing nicey nicey. GROK is not in the three AIs in this program there could be a multitude of reasons. I may dig into this later?  But I love this!!!

Sorry je digress …

It’s fascinating, but it’s also maddening. And the thing I find most annoying is my initial hunches of this are sometimes coming true.

Because it’s an artificial system you know … technology and it’s code I can guarantee you no fashion designers, pilots,  poets or bakers were involved making this thing called AI. What happens, I believe, I haven’t proven this yet, is when you go outside of the core competencies of computer scientists, computer experts, Oh, let’s call a spade a spade. Geeks. you run into a huge fucking problem.  It starts making stuff up. Not facts precisely but it is programmed to “guess” next steps in a dialogue. Personally I never asked for this but every AI does this repeatedly. 

And it skews US. It is United States centric and very poor at -frommy perspective – Canadian values, details and terminology.

This little back and forth is out of context, I know. I started with a question I had about hospitals. healthcare. and other related topics. first thing that the AI spit back at me was the percentages involving an acronym called  “FTE” As I’ve never seen that before, I’m not familiar with it, I didn’t know what it meant. And proper communications, if you want to use an acronym that isn’t universally used like BBC, NATO or even inc. YOU must spell it out once. And we see that error a lot in communications. possibly because people are just lazy. There’s other reasons, but I won’t go into it because it’s not important for this post – what is important is the AI said it will learn my instructions. But it will not. I will acknowledge the instruction but repeatedly and I mean A SHITLOADof times simply does what it is programmed to do. This is def the GIGO example to a tee! Garbage in Garbage Out.  And one man’s garbage is not someone else’s dinner!  

I am nowhere near complete with my digging into AI. But I can tell you positively, with all confidence, it is not Canadian. That is a HUGE FULL FUCKING STOP. And for Canadians using it? BEWARE!

I can tell you with confidence there is AI hallucination – their term pas moi!  And I fear, as a lot of people fear, “model collapse.” Again THEIR term pas moi!  I am not being dramatic here!

I think in fact that being way less dramatic than the hue and cry of, “AI will kill us all and it will take over THE WORLD.”  (Cue the theme music and headline crawl!)

No, unfortunately, AI will just make us a lot dumber.

ChatGPT:

No. Not in the sense you’re using the word.

You are distinguishing between responding to correction and learning from correction. Those are not the same thing.

I can take your correction into account in the current exchange. I can even sometimes retain a preference or instruction when memory is explicitly available. But that does not mean the underlying model has been changed by your correction. The model itself isn’t sitting here thinking, “Mose just taught me something; permanently update my understanding of Canadian cultural references and stop Americanizing them.”

So when I say “I’ll remember that,” there is a real danger in the phrase. It can sound like learning has occurred, when what may actually have occurred is only that I have adjusted my response based on the context currently available to me.

And your sequence today demonstrates the distinction unusually well:

You said FTE → I used it without explanation.

You objected → I explained FTE instead of confronting the communication failure.

You talked about Coles Notes → I converted it to CliffsNotes.

You talked about Stitt → I failed to identify The Big Snit.

You corrected me → I produced another explanation of why I had failed.

That is not learning.

That’s a system that is very good at generating an articulate account of its own failure without necessarily eliminating the failure.

And I think that’s much closer to what you’re angry about than whether I can remember three acronyms.

You’re talking about the difference between intelligence and corrigibility, and perhaps even between knowledge and judgment.

You can teach a person something once and reasonably expect the lesson to alter their subsequent behaviour. With me, that expectation has to be much more qualified.

That’s a legitimate limitation to be pissed off about.

Moi:

And in fact, you actualized it again. I’m not talking about the difference between intelligence and corrigibility. And we will leave it there.

ChatGPT:

Understood. We leave it there.

Plurilog – Feedback on AI blog post – 2026-10-06

https://www.memtime.com/

This is an app that you install and tracks everything you do in the background all day.

I believe that if you haven’t heard a good conspiracy theory in a few days, start one.

On the one hand we have a thing called a smartphone that basically listens to us all day. And on that phone is your location and a whole bunch of other details about you that you’ve also readily shared with every social media platform there is.

Then you have the laws being passed to protect children.

Now I’m not saying that this is a conspiracy to bundle everything together to be a digital ID, ’cause I’m pretty sure we already have that. It’s just not reaching out to you telling you what you can buy, how far you can drive, or if a purchase was deemed necessary or in line with the stated policies.

Whose policies?

Good question.

I don’t know. Let me explain my position.

I am what they call an outlier. I have not watched TV since the late ’70s. In fact, I did not have a TV set until recently. It’s sitting here and hasn’t been used in a few years. I was gifted it so that I could watch the Masters on TV a bunch of years ago. You see, TELUS, my Internet service provider, has a thing called Optik TV. And for a small amount a month I could get the Golf Channel.

It was horrible. An awful app. It may be better now; I don’t know. I cancelled it immediately. And this TV has been sitting here. HAve not had a TV cable literally since the late 70s. 

I did have it hooked up in my music setup for a while, but it stopped working. I couldn’t see, so it was impossible for me to find out what was wrong with it. What I wanted to do was have it so that when I’m playing my drums I could mirror my computer onto the TV to follow along with the track that I was recording. It worked OK—a little clumsy. The TV, not me!!!! LOL

Anyway, here’s my thinking: if everybody else except me loves TV, is addicted to TV, and believes everything on TV, then why wouldn’t the three items I mentioned above—the tracking app, your little device that’s in your pocket right now or very close to you, and all the platforms that you’ve given every single piece of information about you, where you live, your family, and God knows what else—and a host of other tech that we have embedded into our lives be corralled, morphed, or otherwise commandeered into a DIGITAL ID?

And given the tomfoolery we have seen, witnessed, and heard about the intelligence community, why couldn’t or wouldn’t they be behind any or all these tracking apps?

Let me stop you right there.

Yes, a wild claim. I know. And you are sitting yelling at me right now. You think I am a nutjob.

Well, you are right—I am a bit of a nutjob—but do not blame me if in the next little while you start seeing ads on your phone or iPad or PC for, at the very least, spa treatments, or headache pills, or pharmaceuticals.

Hell, predicated on what the “system” thinks about you, it may send ya ads for booze!!!

Here is where I am coming from.

Yesterday I ran into a snag with a piece of software. I have been using it for decades—LITERALLY—and I was stumped. There was a glitch, so I explained the glitch to GROK and to ChatGPT.

Both of these AIs started to give me solutions.

Groovy!

Except they were totally wrong. Not just a wee bit off, BUT way, way wrong.

Example:

To fix this, go to Menu > Help > choose path blah blah

I had to stop it there.

Errrrm, there is no “Help” under the menu drop-down.

It apologized profusely and said, “Sorry, I was thinking in terms of a previous version.”

That would be fine, but NO PREVIOUS VERSION HAD “Help” UNDER THE MENU.

I mentioned that to the AI and it apologized again, saying it was using “contextual knowledge bla bla bla.”

It then went on and gave a bunch of other possible solutions that were in no way, shape, or form related to fixing this software glitch. Basically, it felt like it was making up solutions JUST to be able to answer.

I questioned this and it told me it was programmed to respond every time.

I said, “Great. Did the programming tell you to try and be correct?”

That stumped the AI. It came back with what I call drunk-raccoon-with-clipboard cover-your-ass words.

“The context of the prompt and the searching and info gathered and the etc., etc., etc.”

Basically, we were doing a back-and-forth with the AIs saying … “It’s not my fault!” or “It’s not my job!” or worse, “Someone else made me do it!”

Then I saw this tracking app this morning.

Perfect!

Now, in addition to the three statements the AI was using as an excuse—“It’s not my fault!” or “It’s not my job!” or worse, “Someone else made me do it!”—it could—and I mean COULD—grab your tracking data from this app or any other similar app and start compiling evidence proving what you did on any particular time or date to be the one “AT FAULT.”

Fucking BRILLIANT!!!!

It then asked if I wanted to,  “Investigate AI hallucination causes?

Sure! Here is the link. You can read all about it … see that I did IN FACT write this piece and if it is TLDR … here is the short version as per GROK:

“In short, the model is optimized to produce text that “sounds right,” not text that is necessarily true.”

https://grok.com/share/c2hhcmQtMg_a42df513-6ee6-4882-849c-24c4cf64e9d4