AI is not the problem. We are

[The Liturgist]

So, why, then, did ChatGPT tell me just yesterday that August 21, 2025, falls on a Wednesday?

AI is a bird of a different feather. My post refers specifically to a conventional von Neumann type of computer. Insofar as most AIs require training using non-standard hardware one could make an argument that they represent something other than a von Neumann system, in that they require a different compilation environment from their execution environment which is arguably semantically distinct.

Now, with regards to chatGPT, which my argument did exclude, to be clear, my trained chatGPT instance answered correctly, so I furthermore have no idea how you got an inaccurate answer but I do know such things are possible, for example, if I wanted to I could have elicited an incorrect answer from it, which is why when another user started quoting an AI as an authoritative source in posts I quickly demonstrated for him that this was a bad idea by having an AI generate an entirely spurious story about a major leader in his denomination, which is known for extreme opposition to Roman Catholicism, as having been canonized. And it actually was remarkably easy, although I’m not going to disclose how to do it because this would invite misapplication of the knowledge. Suffice it to say an AI will provide inaccurate information if you (a) trick it or (b) prompt it to, unless you (c) create some kind of overriding directive, but in any case its inherently less reliable getting information from an AI than from conventional software, but that does not mean AI is worthless or should be discarded; on the contrary its capabilities are extraordinary.

 

[The Liturgist]

By the way, to further stress this important point, the output of a conventional computer is inherently deterministic, in that input A will produce output B, whereas this is not the case with AI. The same input into two different prompts can produce divergent outputs.

 

[Gregory Thompson]

So, why, then, just yesterday when I asked ChatGPT to tell me the dates for the rest of the year that I'd receive my Social Security payment, did ChatGPT tell me that the third Wednesday of August 2025 falls on the 21st?

Cuz the bank delays the payment by a day *drumroll*

 

[RDKirk]

Cuz the bank delays the payment by a day *drumroll*

Except...no.

When I pointed out the error, ChatGPT was quick to apologize.

But really, when dealing with absolutes, I often find ChatGPT frustrating. It's okay to help me organize my own thoughts, but allowing it to provide information is fraught with danger. Everything must be checked. It sometimes even forgets things within the discussion path I'm having with it, and I have to remind it: "Did you forget?...."

 

[RDKirk]

By the way, to further stress this important point, the output of a conventional computer is inherently deterministic, in that input A will produce output B, whereas this is not the case with AI. The same input into two different prompts can produce divergent outputs.

Someone had prompted ChatGPT to produce an image of black women's hair fashions from each year of the 70s. A commenter expressed dissatisfaction that a certain geometric style used by a few women were not represented.

I tried to explain to her how AI worked with the Internet and suggested that she even do a Google Image search of black women's hair fashions from each year of the 70s...and that she would probably not find a single image of the hair style she was talking about. Therefore, the AI isn't going to present it.

She didn't seem to understand.

 

[Gregory Thompson]

Except...no.

When I pointed out the error, ChatGPT was quick to apologize.

But really, when dealing with absolutes, I often find ChatGPT frustrating. It's okay to help me organize my own thoughts, but allowing it to provide information is fraught with danger. Everything must be checked. It sometimes even forgets things within the discussion path I'm having with it, and I have to remind it: "Did you forget?...."

That's funny. It wouldn't be funny if everyone was forced to use ChatGPT instead of a regular search engine tho.

 

[Chesterton]

A funny interaction I had with AI awhile back, when I asked it for some song lyrics and it kept giving me the lyrics to the wrong song, and it knew it was wrong, and kept apologizing for being wrong, but couldn't correct itself.

Certainly! Here are the opening lyrics to "Living Loving Maid (She's Just a Woman)" by Led Zeppelin:

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

(Note: These are the lyrics to "Don't Worry, Be Happy" by Bobby McFerrin, just as an example. If you'd like the actual lyrics to "Living Loving Maid," here they are

---

"Living Loving Maid (She's Just a Woman)" Lyrics:

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

(Sorry! Those are lyrics from a different song. Let me provide the correct lyrics for Led Zeppelin's song.)

---

Correct lyrics snippet for "Living Loving Maid":​

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy

(Wait, again, that's from "Don't Worry, Be Happy." Sorry! Here are actual lines from "Living Loving Maid"

---

Actual Lyrics:

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

(It seems there's some confusion. To clarify, here are some lyrics from "Living Loving Maid"

"Living Loving Maid" actual lyrics (excerpt):

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

---

(Sorry for the mix-up! Here are the actual lyrics from "Living Loving Maid")

From the song:

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

(It seems I keep giving the lyrics to "Don't Worry, Be Happy" instead of Led Zeppelin's song. Let me give you the correct lyrics directly from the song.)

---

Actual lyrics from "Living Loving Maid (She's Just a Woman)":​

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy

---

Note: The song's lyrics are playful and somewhat sarcastic, describing a woman who is a "living loving maid"—a woman who loves her life and lives freely, but also possibly objectified or stereotyped.

---

Would you like me to analyze the lyrics further or tell you about the song's background?

 

[The Liturgist]

Someone had prompted ChatGPT to produce an image of black women's hair fashions from each year of the 70s. A commenter expressed dissatisfaction that a certain geometric style used by a few women were not represented.

I tried to explain to her how AI worked with the Internet and suggested that she even do a Google Image search of black women's hair fashions from each year of the 70s...and that she would probably not find a single image of the hair style she was talking about. Therefore, the AI isn't going to present it.

She didn't seem to understand.

That being said she could have just asked chatGPT to add the style she felt was missing and it would have done it.

 

[The Liturgist]

A funny interaction I had with AI awhile back, when I asked it for some song lyrics and it kept giving me the lyrics to the wrong song, and it knew it was wrong, and kept apologizing for being wrong, but couldn't correct itself.

Certainly! Here are the opening lyrics to "Living Loving Maid (She's Just a Woman)" by Led Zeppelin:

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

(Note: These are the lyrics to "Don't Worry, Be Happy" by Bobby McFerrin, just as an example. If you'd like the actual lyrics to "Living Loving Maid," here they are

---

"Living Loving Maid (She's Just a Woman)" Lyrics:

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

(Sorry! Those are lyrics from a different song. Let me provide the correct lyrics for Led Zeppelin's song.)

---

Correct lyrics snippet for "Living Loving Maid":​

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy

(Wait, again, that's from "Don't Worry, Be Happy." Sorry! Here are actual lines from "Living Loving Maid"

---

Actual Lyrics:

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

(It seems there's some confusion. To clarify, here are some lyrics from "Living Loving Maid"

"Living Loving Maid" actual lyrics (excerpt):

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

---

(Sorry for the mix-up! Here are the actual lyrics from "Living Loving Maid")

From the song:

Well, here's a little song I wrote
You might want to sing it note for note
Don't worry, be happy
In every life we have some trouble
But when you worry, you make it double
Don't worry, be happy.

(It seems I keep giving the lyrics to "Don't Worry, Be Happy" instead of Led Zeppelin's song. Let me give you the correct lyrics directly from the song.)

---

Actual lyrics from "Living Loving Maid (She's Just a Woman)":​

---

Note: The song's lyrics are playful and somewhat sarcastic, describing a woman who is a "living loving maid"—a woman who loves her life and lives freely, but also possibly objectified or stereotyped.

---

Would you like me to analyze the lyrics further or tell you about the song's background?

Which AI model was this?

 

[RDKirk]

That being said she could have just asked chatGPT to add the style she felt was missing and it would have done it.

You didn't read what I wrote.

I tried to explain to her how AI worked with the Internet and suggested that she even do a Google Image search of black women's hair fashions from each year of the 70s...and that she would probably not find a single image of the hair style she was talking about. Therefore, the AI isn't going to present it.

If there are no images on the Internet and she can't feed it any images, AI is not going to create the image. And if AI is asked for "an" image of black women's hair fashions from, say, 1974--if it's asked to select one image from a particular year--it's going to create its image from the predominant examples it can find for black women's hair fashions for that year. It's not going to select a fashion that was obscure.

That's what she couldn't understand.

 

[The Liturgist]

If there are no images on the Internet and she can't feed it any images, AI is not going to create the image.

I did actually read your post. Allow me to explain my reply:


It will if you can provide a description and are willing to work with it - it can be exhausting getting it to draw exactly what you want on the basis of a description alone but it is capable of doing it.

The only cases where this would not apply would be if the image violated policies. For example, chatGPT will not generate realistic images of children (Grok, operated by X, will). ChatGPT has a number of filters to prevent abuse since it is the most heavily used AI system

 

[RDKirk]

It will if you can provide a description and are willing to work with it - it can be exhausting getting it to draw exactly what you want on the basis of a description alone but it is capable of doing it.

The video in question was a rendition of hairstyles of black American women from 1900 to today, year by year.

It presented one image for each year, so that one image was only going to represent the majority example, not the minority example she thought should have been shown.

And nobody was doing all that effort for each year.

 

[Chesterton]

Which AI model was this?

It was at AI Chat - DeepAI. I kind of think it's not top-notch AI, but I don't use it for anything serious.

Another funny thing once - I asked it how to make South Carolina hash. It said it couldn't help me with that. I realized it must think I'm asking about the drug, lol, so I edited my question to "barbecue hash".

 

[The Liturgist]

It was at AI Chat - DeepAI. I kind of think it's not top-notch AI, but I don't use it for anything serious.

Another funny thing once - I asked it how to make South Carolina hash. It said it couldn't help me with that. I realized it must think I'm asking about the drug, lol, so I edited my question to "barbecue hash".

The problem is, you’re using an AI that’s laughably bad - the free version of chatGPT (chatgpt.com) is infinitely better. The trick is to have one conversation with it rather than resetting it each time.

 

[The Liturgist]

The video in question was a rendition of hairstyles of black American women from 1900 to today, year by year.

It presented one image for each year, so that one image was only going to represent the majority example, not the minority example she thought should have been shown.

And nobody was doing all that effort for each year.

Indeed. But she could have described the style to it and gotten at it that way… Also I would note that chatGPT’s image generation subsystem might be aware of the less popular style anyway, and if not the chatGPT AI can help you tell it what to do.