Preachers and AI

[JohnH]

I used no AI at all when I wrote that message, or any of my other messages in this thread.

Just pure, unadulterated Bootstrap in those messages.

In that case, we now have two forum participants here who tend to write like LLMs do, which is certainly unusual but within the realm of possibility.

Maybe next we'll have a preacher who talks like an LLM, too. I have wondered for a while if reading a lot of LLM output eventually causes someone to write / speak like they do as well.

Nonsense. Some people really do examine things from many sides, regularly, habitually. Not using the stereotyped LLM formulation you mentioned, and I did not. Examining various sides of something and various possibilities is a regular part of thinking things through.

I think you will find that I have always done this if you look through the archives.

Do you not agree that having a final paragraph or sentence of “It’s not X. It’s Y” is a common LLM pattern?

[ken_sylvania]

Did you naturally choose to use the "It's not X. It's Y" phrasing here, or did the AI suggest that?

Here's the first message in this thread where I used AI at all. Following Ken's lead ... do you think these texts are AI-generated?

* Hosea 6:6
“I desire mercy, not sacrifice, and acknowledgment of God rather than burnt offerings.”
* 1 Samuel 16:7
“People look at the outward appearance, but the LORD looks at the heart.”
* Matthew 9:13 (quoting Hosea)
“I desire mercy, not sacrifice.”
* Matthew 12:7
“I desire mercy, not sacrifice.”
* Matthew 23:23
“You have neglected the weightier matters of the law: justice and mercy and faithfulness.”
* John 6:27
“Do not work for food that spoils, but for food that endures to eternal life.”
* John 6:63
“The flesh counts for nothing; the Spirit gives life.”
* Romans 14:17
“The kingdom of God is not eating and drinking, but righteousness and peace and joy in the Holy Spirit.”
* 1 Corinthians 4:20
“The kingdom of God is not a matter of talk but of power.”
* 2 Corinthians 3:6
“The letter kills, but the Spirit gives life.”
* 1 Peter 3:3–4
“Your beauty should not come from outward adornment… Rather, it should be that of your inner self.”
* Mark 7:15
“Nothing outside a person can defile them by going into them. Rather, it is what comes out of a person that defiles them.”

Those quotes don't use the "It's not about ...., it's about ...." phrasing that JohnH was talking about.

My opinion is that the reason AI likes that formulation so well is because it scraped so many examples of it off of the web from humans writing it.

[ken_sylvania]

Do you not agree that having a final paragraph or sentence of “It’s not X. It’s Y” is a common LLM pattern?

It might be a common LLM pattern, but it doesn't necessarily follow that the vast majority of uses of that formulation are LLM inspired.

[JohnH]

I used no AI at all when I wrote that message, or any of my other messages in this thread.

Just pure, unadulterated Bootstrap in those messages.

In that case, we now have two forum participants here who tend to write like LLMs do, which is certainly unusual but within the realm of possibility.

Maybe next we'll have a preacher who talks like an LLM, too. I have wondered for a while if reading a lot of LLM output eventually causes someone to write / speak like they do as well.

You think this is an LLM generated song or what?

It's not just about the manger
Where the baby lay
It's not all about the angels
Who sang for him that day
It's not all about the shepherds
Or the bright and shining star
It's not all about the wise men
Who travelled from afar

It's about the cross
It's about my sin
It's about how Jesus came to be born once
So that we could be born again
It's about the stone
That was rolled away
So that you and I could have real life someday
It's about the cross

It's not just about the good things
In this life I've done
It's not all about the treasures
Or the trophies that I've won
It's not about the righteousness
That I've find within
It's all about His precious blood
That saves me from my sin

It's about the cross
It's about my sin
It's about how Jesus came to be born once
So that we could be born again
It's about the stone
That was rolled away
So that you and I could have real life someday
It's about the cross

The beginning of the story
Is wonderful and great
But it's the ending that can save you
And that's why we celebrate

It's about the cross
It's about my sin
It's about how Jesus came to be born once
So that we could be born again
It's about God's love
Nailed to a tree
It's about how every drop of blood
That flowed from Him when it should have been me
It's about the stone
That was rolled away
So that you and I could have real life someday
So that you and I could have real life someday
It's about the cross (it's about the cross)
It's about the cross
It's about the cross
It's about the cross

Nothing there matches the specific way LLMs like to use a “It isn’t X. It’s Y” pattern as their final sentence to conclude a codex of generated text.

It turns out this is well studied (I never read other sources on it before; it’s just something I noticed myself):

https://www.theguardian.com/commentisfr ... ot-x-its-y

[JohnH]

Do you not agree that having a final paragraph or sentence of “It’s not X. It’s Y” is a common LLM pattern?

It might be a common LLM pattern, but it doesn't necessarily follow that the vast majority of uses of that formulation are LLM inspired.

Well, except it was very rare to see the GPT style formulation at all before LLMs became popular. That leads me to reason someone is either using an LLM to generate text, or has picked up on the style of LLMs and parrots it.

[Bootstrap]

Those quotes don't use the "It's not about ...., it's about ...." phrasing that JohnH was talking about.

I didn't use the precise "the specific way LLMs like to use a “It isn’t X. It’s Y” pattern" either. The "just" is actually significant here. I don't know if LLMs use that or not.

It's not just about whether you use AI or Google or Wikipedia or an encyclopedia or whatever. It's about how engaged your mind is, whether you are educating yourself or looking for an easy out. You can use AI to get smarter or dumber, and which you choose is up to you.

In a lot of LLM output, what's really annoying is that the "It's not about" clause is often irrelevant, it complicates the line of thought needlessly, and making it harder to see what it IS about. It's overused.

My opinion is that the reason AI likes that formulation so well is because it scraped so many examples of it off of the web from humans writing it.

I'm sure that's true.

[Bootstrap]

Do you not agree that having a final paragraph or sentence of “It’s not X. It’s Y” is a common LLM pattern?

It might be a common LLM pattern, but it doesn't necessarily follow that the vast majority of uses of that formulation are LLM inspired.

Well, except it was very rare to see the GPT style formulation at all before LLMs became popular. That leads me to reason someone is either using an LLM to generate text, or has picked up on the style of LLMs and parrots it.

OK, how is it that you are such an instant expert on English usage and my own usage of English? Document your claims please, carefully, in another thread.

Fact is, you were as confidently wrong as any LLM in your pronouncements. And we can decide to be dumber or smarter with broad stereotypical pronouncements too. Can you just admit you were wrong? Or actually do the vast amount of work needed to demonstrate your claims are true?

[JohnH]

Those quotes don't use the "It's not about ...., it's about ...." phrasing that JohnH was talking about.

I didn't use the precise "the specific way LLMs like to use a “It isn’t X. It’s Y” pattern" either. The "just" is actually significant here. I don't know if LLMs use that or not.

A comom term to refer to this is “It’s not just X, it’s Y” (although an LLM generally won’t make a run on sentence like that). For example: https www.reddit.com/r/ChatGPT/comments/1l8ha ... t_x_its_y/ (caution: strong language)

Given I believe you are well educated about AI LLM tools, I am surprised you are unfamiliar with this.

[ken_sylvania]

Those quotes don't use the "It's not about ...., it's about ...." phrasing that JohnH was talking about.

I didn't use the precise "the specific way LLMs like to use a “It isn’t X. It’s Y” pattern" either. The "just" is actually significant here. I don't know if LLMs use that or not.

How can you know if you used the LLM pattern if you don't know whether or not your pattern is the type LLM uses???

[JohnH]

It might be a common LLM pattern, but it doesn't necessarily follow that the vast majority of uses of that formulation are LLM inspired.

Well, except it was very rare to see the GPT style formulation at all before LLMs became popular. That leads me to reason someone is either using an LLM to generate text, or has picked up on the style of LLMs and parrots it.

OK, how is it that you are such an instant expert on English usage and my own usage of English? Document your claims please, carefully, in another thread.

Fact is, you were as confidently wrong as any LLM in your pronouncements. And we can decide to be dumber or smarter with broad stereotypical pronouncements too. Can you just admit you were wrong? Or actually do the vast amount of work needed to demonstrate your claims are true?

Boot, I’m saying this nicely as a friend: your writing is far more LLM like than it was pre-2025. I don’t know why this is. It’s just an observation.

If you really want to do a careful analysis, I have 3 codex resets left expiring in a week or two which should be more than enough to scrape your posts on this forum and run a comprehensive analysis of changes in writing style including LLM style tells. Of course that won’t tell me you’re using an LLM - it just says your writing is resembling a certain pattern.

My brain is also a pattern recogniser, so I recognise when a writing style changes. You aren’t the only person this is happening to. This is a well documented phenomenon; this article cites a number of high quality studies:

https://www.axios.com/2026/05/02/ai-cha ... g-speaking

I’m sure my own style has changed as well. Language changes over time. But the AI field and LLMs also change too, and I tend to have a sharp eye and ear for LLM generated text since I treat it as having far less value than human-produced language.