AI Revisited: The Turd in the Digital Pool

The Honeymoon Phase is Over.

This week, Josh and Kris revisit their stance on Generative AI. While early adoption promised a reduction in cognitive load for DMs, the reality has shifted toward a "net negative" across nearly every metric. From the environmental impact of data centers to the "enshittification" of search results, the hosts break down why the "word guessing machine" is failing the people it promised to help.

Topics covered include:

• The Admission: Kris publicly admits Josh was right about the "AI bubble".

• The Tech Cycle: Late adoption vs. the "new toy" syndrome.

• Art Theft: Evaluating the "British Museum" argument for AI training.

• Linguistic Limits: Why English is a "hoarder language" and how "burstiness" keeps AI detectors effective.

• The DM’s Craft: Why the "lonely fun" of manual world-building remains an essential, deeply human artistic act.

Transcript

You're listening to No Plot Only Lore, a podcast about games and the tables we play them at. Your DMs tonight and every

night are Josh and Kris. You can find us on all podcast platforms or check us out at noplotonlore.com.

If you like what you hear today, please rate and review the show and share it with everyone you've ever met. Welcome

back to No Plotton Only Lore. This week brought to you by Chat GPT. The AI that

was meant to revolutionize the world that sucked so bad we've got to sell ads on it now.

I'm so happy they're not making money and this thing might implode within the next year.

It's so incredibly cathartic to me that their

art theft sucks that bad. I this is like very much the ultimate

shot in Florida for me. Like I'm so glad I was right and they are bad. I like from minute one I was like this sucks

and is stupid. And I think as far as I can tell, the only reason

that it's continued to trendle along was one the promise of like well they're not

saying it but the promise of like hey we can just like trick this thing into working for free and then the fact that

like without this AI bubble the American economy would actually be in like a full-blown depression.

Yeah, it is the only thing keeping us afloat right now and that is horrifying on so many goddamn levels.

Um, it's awful. So, for our listeners at home and very

publicly and with humility, I say Josh

was right about everything and I was incredibly wrong about AI. Godamn.

This feels so good. I the fact that it took the fact that it took a year to hear that out of you really just like

just sweetens it that much more. Um I mean that's the first time you've heard it publicly. I've I've said it a

couple of times with the the mics off, but like yeah. Oh my god. I

I have a history of early adoption on things. Like I I like new stuff. I I

like to play with new toys. I I was one of the first people to pick up like a new iPod when they turned to like the

iPod touch. I was in line to grab an iPad when they first came out. I was lined up outside of a drugstore in

Victoria on my birthday to buy a Switch, right? Like I I like new stuff. I like

playing with new things. And AI, man, they sold it good.

I I am very much the opposite of you in that regard. I buy things on the third

generation. I I hung on to my 80 gig iPod classic until they tore it from my

hands and I could have an iPhone big enough to put all of my music onto it because there was no way I was leaving

the house without my entire music collection every time. Um I Were you a Zoom kid? Did you have a

Zoom? Oh, I was worse than a Zoom kid. I was a minisk kid. I I

Yeah, because for some reason I I trusted like physical media more than

like the idea of like it's all there on a little hard drive. I don't know why, but yeah, part of it definitely was like

just budget. Like I just couldn't buy one of them and I had to spend my own money on it. So yeah, I I

was a late adopter and then I hung on forever. But that's how I I wait for everything to go through the first three

or four iterations of it's bad, it's bad, we're fixing it. Okay, now we

proven it's got longevity and we've got all the bugs worked out, right? So

as in any sort of tech, that's the way I go. Even even like I got an N64 in like

1998. Like it was way late in the cycle. It was like I bought an N64, they

released Bon Conquer Bad Fur Day and killed it. I'm like, "Oh, well, all right. I guess I'll go pick up all the games that are good." Like,

Conquer Bad Fay was such a great [ __ ] game, though. Yeah. Okay. I spent so much goddamn time

playing the like Squirrels versus Teddy Bears nonsense, like multiplayer on that [ __ ] So many

hours of my life have been devoted to that game. See, I I played um I everyone plays

Golden Eye. We played a lot of Tony Hawk, but I never got into like the the Perfect Darks. I played Perfect Dark a

bit, but not like on my system. What my brothers and I spent a lot of time playing was Battle Tanks for some

reason. Like there's always everyone's got that one weird game that like hit

their family at the right time and that became your multiplayer. Like yeah, we played a little Mario we sorry we didn't

play Mario Kart, we played Diddy Kong Racing, which was better Mario Kart. Um so much better planes. planes,

hovercrafts. Yeah, it was awesome. And then yeah, Battle Tanks, and um I spent

probably a decade hunting for a copy of uh Mace: The Dark Age, which was like

this weird fighting game that I had played on an arcade at Laser Tag one

year, did really well at it, and was like, I this is the best game. I got to get it. spent 10 to 15 years trying to

find a copy that half worked and it I was like, "Oh, this game is awful. This is this is terrible. Why why did I

obsess about this?" My mom kind of did the same thing. She got a a Super Nintendo with her um tax

credit the year it came out. And clear that was her Super Nintendo. We got to

play it when she wasn't. And we rented a game from a Blockbuster.

God, that's a weird sentence to say now. Yep. Um, and it was called Brainlord, and it

was like a top-own RPG that involved a lot of like puzzle solving, and she became absolutely obsessed with this

game, and we needed to own it, and she couldn't find a copy forever until the Blockbuster started selling their SNES

games, and then she bought the version that we rented. Like yeah, I

we Yeah, we we rent like when I think about renting video games on New Year's

Eve in like 2000, maybe 2001, uh me and a couple of

buddies from junior high rented an Xbox from Rogers Video and stayed up all

night eating chocolate covered coffee beans and playing Halo. God damn. Yeah,

them were the days. That was probably the best New Year's Eve of my life. The next day was just misery. But

anyways, yeah, late adopter on tech, late I don't want to say I'm a skeptic. I'm just in in this aspect conservative.

I know the first generation of everything is the most expensive, the most fragile, the least, you know,

functional. Um, and so I wait. I I never got on 3D TV. And I was right. I didn't

touch I didn't touch like any HD TV stuff until Blu-ray had one and was

inside of my DVD player on my Xbox, right? Like I just right

I don't go for the new flash because I think it's stupid. Uh the same with um

uh virtual reality. Uh my my son has a VR headset that he bought at his mom's

house. I haven't touched it and I never will cuz it's clearly dead now. Like I just don't care, right?

Yeah. So, it's like the second time they've done that, too. Like VR seems to go in like weird cycles of

making people hurl. Um I don't have that. But like I

maybe it's just like a financial thing. I'm like I'm not going to waste my money chasing these fads because there's always a new thing to chase that they're

going to sell you on. But did I ever tell you about the fact that I used to get virtual sickness from

first person anything? No. Yeah. So, when I was uh playing

first-person games of any kind, if I could see it moving on the screen, and it should be moving past my eyes, but it

didn't, it would make me hurl. Oh, that's wild. And before my son was born, I got myself

a bucket and a controller and Minecraft because it seemed like it would be the like least likely to cause that

response. Yeah. and just had to train myself to do it because I didn't want my kid to be

the one like the weird kid that didn't have Minecraft. That's wild. But like even other people playing those

games Yeah. Um if there wasn't like a a character for me to focus on in the

middle of the screen Yeah. it would cause me to hurl. That's crazy. Nauseous. Like some of my earliest gaming memories

are playing like Wolfenstein and Doom and stuff like that. Like just never had an issue with first-person games. Even

like um Descent was one that had like crazy radical full 3D motion. There was

an Alien versus Predator game where if you played as the alien, literally every surface was traversible, so you could

just like spin your way down a hallway coming after Yeah. Yeah. Mine was bad

enough that I tried to play Mass Effect in third person. Gross.

It did not go well. No. And because of that, I'm also really bad

at first person games. Anyway, uh yeah, video games aside, AI played them for me.

E. Okay, so you've admitted that you were wrong and it was a mistake. Why?

What was the trigger for you that what said, you know what, never mind? Okay. So,

I understand logically and morally

that AI has always been art theft, right? It it is a crime against art and

humanity and it always has been. And I'm not

as opposed to that as some other of my peers who have like taken ethics and

like dealt with art like aesthetic, right? Like it's awful.

But sometimes art theft is progress, right? Like there have been a lot of occasions where like just straight up

jacking things has led to interesting scenarios. When it's worth it,

it isn't okay. It It is still an evil, but when it creates great outcomes for

humanity, it can sometimes be considered worth it in the long run. Okay. But like, how are you defining art

theft? Like, are you just saying like all mimicry is theft? No, I'm saying that the wholesale use of

original artistic work to train AI through

literal theft. No, that I understand. I'm talking about like in the past where you're like,

well, sometimes art theft was good. Not good. Not good, but understandable.

Okay. Right. Like what one of the easiest examples would be the entirety of the British Museum.

Okay. It's just art theft, right? Like the entire thing is art that was stolen from

other places. Sure. Right. They are protected. They are

preserved. Is it a good thing? No. Other countries that had their stuff taken

should probably have their stuff back, right? But right, but then there's the argument

that like without a a system like the British Museum, a lot of this stuff would have been lost to history or

destroyed through regime changes, etc., etc. I mean, like look at what happened with ISIS and their destruction of, you

know, massive historical monuments for ideological reasons. You know, having

it's one of those ends justify the means kind of arguments. At the end of the day, is it the right thing to do? Not

always. Are the results Almost never. Well, I mean,

okay, if you're going to raise a culture to the ground, stealing some of its

treasures to show off to the future is probably the right thing to do as opposed to eras.

Tuck some stuff away and then light all the people on fire. Yeah, exactly. If you're already going to be an

[ __ ] then maybe this is a way that you can be an [ __ ] that preserves something a little bit,

right? Anyway, art theft always evil, not always

terrible, I guess, right? Like not always the worst possible thing.

The problem that I have with generative AI is that it is not good at the things

that it purports to be good at. Correct.

If it were good at stuff, if it were good at stuff that was going to

materially improve the lives of the majority of the human race

like it was pretending it was going to do, then okay. But instead, it has been a net negative

in almost every perceivable way. on every metric that we have. It has been an absolute [ __ ] show,

right? It's bad at stuff. It uses a huge amount of water. RAM prices are through the

roof. You can't buy a graphics card anymore. Mhm. We've got data centers that are creating

huge amounts of pollution, both like noise pollution and air pollution and water pollution that are cropping up

everywhere. They're propping up the entirety of the American economy right now in a way that is probably unhealthy

and going to lead to long-term greater suffering. Yep. It's like every metric I can think of.

Yep. Putting huge extra strain on power grids. Um just just the fact that they take up so

much space and then leave that area just sort of

this desolate little block. Yeah, they're awful. And like it's impossible to live in those areas in a

lot of cases because the the noise pollution has gotten so bad and like the the pollution pollution has

gotten so bad that the people who were living there before the data center went up have to move.

Yeah. Right. displace people which like

again if it had done the thing it was promising to do

right and the thing that it was promising to do was improve human lives

somehow they were always really vague about that they're still vague about it I mean they're still they're so vague about it

that they're going to be selling ad space well I mean what's his name from Microsoft there Sasha Nadella has been

like, "Hey, we just got to like keep funneling money down this hole until someone figures out how to make it

beneficial for humanity." Like I that's

I'm not a pessimistic person by nature, but that is too much optimism. And it's

optimism fueled by I feel like just a a massive version of a sunk cost fallacy.

Like, yep, we keep pouring money into the infinite money hole. Eventually, eventually we

got to see the bottom start to come up. Right. Well, like sunk sunk sunk cost fallacy

plus the gamblers's fallacy. Right. Right. We keep rolling seven.

Yeah. Right. We keep We keep betting seven. We keep rolling one. But someday we got to roll a seven and we're going to win big. We're going

to win ball back. We just got to push. We just got to push. We're breaking even. Yeah. Yeah. We're doing good.

No, it's Yeah, it And part of it too is that like they took the initial promise

and I think that like when we did our first episode Yeah. everything about AI up until that point

was still opt in. Right. Right. You had to go looking for it.

Yeah. It was like, hey, there's this tool over here. You can go poke around with it if you want. And you know, it

was very exploratory and and like a fun toy in the corner. And now it's like my

your coffee maker has AI. It's like what? Like there's no escaping it.

Yeah. Like the every web app that I use and I use a lot of web apps across my day.

Yeah. Except the ones that we've abandoned at work. Yeah. Yeah.

The only one that I can think of that does not use that is one that I am the only person who's qualified to handle because it is so old that nobody's

learned how to use it. Yeah. Like that's the only thing that doesn't have AI stuffed into it. And I think as

soon as our bosses realize there isn't AI in it, they're going to put it in. Yeah. Yeah. They're going to find a way

to wedge it in there even though it hasn't been touched in years. Like it's everywhere. It is inescapable.

Like what one of the things that really tripped me on this was I went to go and

do a Google search. Yeah. And for a long time almost my entire

like online experience searching on Google had Google search or search Google or I'm feeling lucky or

whatever the the phrasing was around the bar. Now it had ask Google,

right? Because the thing that Google is going to give me is not search results, which hasn't been true for a while.

But sure, it's going to give me an AI summary of the results, right? Yeah. Yeah. Yeah. It's not Oh,

nothing makes me angrier than seeing the AI summary at the top and knowing for a fact that it's not accurate. Like,

yeah, I I spent a lot of time getting very good at my Google food, you know? um

their their spiders worked in a particular way, their algorithm worked in a particular way and if you knew

boolean searching, it was extremely accurate and you were able to dive

through that information to get what you wanted. And now they've obuscated the results in favor of trying to interpret

your intent, which is miserable because they want to be able to provide you better ads.

Right. Right. It's It's made my search results demonstrabably worse because I'm

trying to use the skills that I had to get what I want and now

the AI chatbot doesn't know how to read what I asked it.

So, okay, I have been like deep in the guts of the internet

for a long time. Yeah. I'm not good at it. I'm not a programmer by nature. Yeah. But I have like spent a lot of

time on like Favia's search lures sites. I have spent some time developing my own spiders. I have like done a bunch of

like not necessarily over the table searching in the past.

So when I came across this like ask Google thing, my first response was maybe I should just make my own spider.

Yeah. And then I remembered that my spider is just going to find AI slob.

Yep. Yeah. Because not only has it stolen all of the information out of

there to try to like do what it wants to do with it, but it's now then turning around and polluting the original source

of it with its misinformation, hallucination, um

just fantasy land results. Well, and that adds to the problem that

it was originally supposed to solve, right? Like

in our first episode about AI, which was admittedly very early in the noplot only lore URA,

the thing that I was purporting to be like

impressed by for this was the reduction in cognitive load for DMs, right? Sure. You didn't have to spend a lot of

time putting together a stat block. You could tell the AI what type of stat

block you were looking at for what type of stats you would like to have it like appear in the list.

Mhm. And then it would give those to you. It was great, right? Like I didn't have to do any of

the math. I didn't have to do any of the scaling. If I wanted to like level up a creature to a like different challenge

rating, then it would do all of that math for me. [ __ ] great. But now

the amount of extra factchecking, correcting, reprompting, retooling my

prompts, like prompt engineering just Yeah. Oh my

all of that and like managing the output and trying to like think about all of

the different like iterations of the thing that I can ask for and like how can I tweak it. A lot of times at this

point it's gotten to the point where it would just be faster and easier for me to [ __ ] do it myself.

Yes. Yeah. I think I think the biggest problem is that like when it was early and getting things

wrong more often those results and that stuff was not

like sanitized, right? It was just we dropped the turd back in the pool. Now

take another sip. Um gross. Well, yeah, but like it just

and shitification very literally. Yeah, for real. Like it it

if it had time where it went through this process of like people verifying the results and be like, "Yes, that's

that's accurate. That's how that works." You know, it doesn't Okay, if you built

a little tool that took the scaling rules from fifth edition and you could

just like drag and drop whatever creature you wanted and then hit a level

and then it would [ __ ] it back out, you would know that it would work 100% of the time because it was just math,

right? Yeah. And it turns out AI can't do math and it can only kind of half understand

your questions. So it just kind of like it's it's what did someone call the

other day? Someone called it a word guessing machine and and

that is so remarkably accurate. Yeah. Um one of the things that I have been

Okay, so like we we work with AI at our job. Oh, I was gonna say at gunpoint. I

Yeah, at gunpoint. Like neither of us is super keen on the fact that AI is a part of our job.

We have we have been told like in one of the feedback rounds where it was like, "Hey, tell us how you used

AI to make your job better after just like dumping AI on us." I'm like, "We

don't I don't use it at all. It has not had any effect on me." And they're like, "No, you have to by this time next year,

you have to tell me how you've used AI to make your job better." Which is like just a terrifying way to handle that.

Yep. Yep. Um, but like because we do use it at work, I

am still intrigued in finding ways to make it better, right? To make it like

actually do the things that it's supposed to do. And

one of the projects that I was working on with it for a while is a weight loss project. And so every day I would put my

current weight, my current calories into the AI. I was using chat GPT5 at

the time, and I would have it occasionally do some like math on those. There was like a

single chat that I was using for all of it. Um, and it would sometimes make mistakes

that if I were a different person could have been fatal, right?

Right. Like it was trying to have me eat a lower amount of calories every day.

Yeah. Right. It was trying to make me reach like um record lows on my like record

highs on my deficit, record lows on my calories, which if I'm the type of person that's

prone to an eating disorder, all that does is encourage me to eat

less because it had no context or frame of reference for like

what does a calorie mean and how many does a human being need to, you know, function and and maintain weight rather

than just greater into the skeleton. Well, and the problem is that I'd given it that context already, right? Like one

of the ways that chat GPT and other large language models

work is that when you send them a message, it isn't just sending them a single message. It is sending them the

entire history of that chat. everything that they've said, everything that you've said, and then it reads that,

contextualizes it as much as it can, and then spits out a response from there.

Sure. But which is part of why it has this huge environmental effect is because every conversation is actually 50

conversations. Yeah. And part of the problem is that AI

isn't necessarily great at contextualizing a huge amount of information all at once, right? Like it it's again one of the things that it's

purported to be good at is handling huge amounts of data but when it is given

that huge amount of data it starts to forget things right. So it starts to forget.

It's baffling to me that the machine forgets. It never remembers

that's the other thing. For so long the problem in the tech world and with computers was like hey

they they do exactly what you tell them to do. they can't forget things, right? Like like they hang on to all of this

information and then if you tell it the wrong thing, it's because it's remembering everything and trying to do everything at the same time. And now we

have the opposite problem. As a person who works in tech support, there is nothing sadder to me than the

loss of the phrase, "The computer only does what you tell it to do." Yeah. Yeah. 100%.

That isn't true anymore. Like that was my favorite way to tell them

that the problem existed between the keyboard and the chair. Like Yeah. Yeah. You you called in, you had

an ID10T problem and then you just you're like, "Well, call back when you follow my instructions." Exactly. Like

Yeah. Yeah. Um No, it's it's really brutal. And the so yeah what was happening for my weight loss project is

that I had created too much history and it began to summarize that history in

its own processing. Yeah. And so it was pulling information selectively that when processed in the

way that it was processing it could be dangerous. Right. Well yeah I that's the number one

feature I see right now on things is oh well let AI summarize this for you. I'm like, "Don't you goddamn dare." Cuz it

you're so bad at it. It's so bad at it. It loses context. It loses reference. It just skips things.

Because there was there was a point in time where we were using uh well, we still are, but there's

there's an AI feature that summarizes a call that comes in, and for whatever

reason, arbitrarily capped itself at five lines. So, it was just kind of like decide what information from an hourong

phone call was the most important to fit in five lines of context and just like it's not that it had the capacity

to condense an hourong phone call into five lines of content. It just chose

five sentences out of the hourong phone call. And I feel like it doesn't even

everyone and not even like the the five most relevant sentences necessarily, just the

five sentences that AI thinks were the most important either because we spent the most time on them, right?

Or because they stood out as being like sufficiently bursty,

which is like another thing that I find really interesting about the like AI phenomenon.

Yeah. Is that like so much of it is based on linguistics.

Yeah. So, like one of the projects that I was doing recently, one of the things that I was trying to sort out was like

how to get AI to jailbreak the AI detectors.

Okay? Right? Like I wanted AI to be able to produce text that seemed like it hadn't been created by an AI,

right? And so I was giving it very specific instructions for how to get through that

by like increasing the uh variation in paragraph sizes, increasing the

variation in word selection. Um increasing um

it's called burstiness, which is like the the difference in sentence length that okay you're dealing with. And so

like giving it very specific instructions and every single time the AI um detectors were coming up this was

like 99% AI written like 100% most of the time 99% was the best I got it down

to. That's incredible. And it was because even with those very specific instructions, it can't go past

that programming. Yeah. It has these like very clearcut

uh like instructions that it has to go by based on

what we think an average should look like. Right. Right. Right. Right. Well, what's

very funny to me is that like because you complained earlier this week that uh

about English specifically as a trash language and didn't complain, celebrate it.

Sure. Um, English is a trash pile of a language. It is the hoarder language. It is

stealing vocabulary from other languages and stuffing it into hoarder piles in ways that make absolutely no goddamn

sense. And I love and appreciate that about this language so goddamn much. Well, and what's very funny to me is

that it's those very features of the English language that are

helping us to keep AI at bay. Because

AI craves uniformity, right? And human

beings, especially English-speaking human beings, are just not set up that way. Um,

not even a little bit. the way that each of us, you know, the the fact that you

can say that a writer has a distinct voice when using the exact same set of

tools that every other person who speaks this language has is so funny to me. Like the there's other languages out

there with much more strict and formalized structure and and you know, verbiage and things like that and

English just says do whatever you want as long as someone gets the idea. Uh-huh.

Which again is one of my favorite things about this language. Right. Like the only rule is

transmission, right? As long as the thing that I am saying becomes something that you understand,

we have done our job as a language. See, and that's that's where you there's one part where you and I strongly differ

and I think it's uh in the importance of spelling. Um

and my is so new to my end. It's exactly that importance

on um the communication and the the

clarity of what you're saying that is the reason that spelling needs to be there. No, I understand the the drive for

standardized spelling and grammar. I understand why we want standardized spelling and grammar, right? I also recognized that up until

about 250 years ago, they didn't [ __ ] matter, right? But 250 years ago, we also didn't

have cars. Like, language is a technology that can improve over time. And I think spelling and grammar are new

additions to that technology that we need to treasure in the same way that we do seat belts and windshield wipers.

Hell no. taking the seatelt off, hitting the gas pedal as hard as I can, straight

into like runon sentences with no punctuation. Like we're we're [ __ ] doing this. Well, and that's that's what's funny is

that that instinct and that lack of structure from people in general is what

often helps like differentiate their like human writing from chat GPT and

other AI generated language. So one of the things that chat GPT does is and

this is the the part of the like word guessing game thing that you were talking about is that as it's

constructing a sentence the thing that it's doing is doing math to see what

would be the next word statistically speaking. Yeah. Yeah. And it can't break out of that. It does

not have the ability to break out of that in normal circumstances like unless you jailbreak chatgpt itself. right?

Which is possible and there's a bunch of like fun things that I've watched people do for that. But like unless you break

chat GPT itself, that math always maths, right? And that's the thing that the AI

detectors are looking for is like, is this word statistically the most likely to come after the word before it?

How many times do you deviate from that? You know what this makes me think of is when they are trying to determine

whether or not someone's been cheating at chess. where is is the next move the mathematically correct move to make and

then how many of those in a row does it do? Um if you're constantly the best move,

you're cheating, right? Um and again compared to your ELO, right? Like if Magnus

is right doing the best moves. Yes. Then he's doing the best moves.

Well, like Yeah. Like Magnus Magnus Carlson is going to make the correct move 90% of the time. If it jumps to

97%, that's odd. Cheating, right? But the average person might only

do it like 50 or 60% of the time. All they have to do is jump to 75 and it's like, "Oh, they're they're cheating or

they're getting coached." Magnus Carlson's opponent is doing statistically better than typical. There

might be a vibrator in his ass. That's a whole other story. Um,

I was going to say it it does bring up, not that I like want to explore the possibilities of AI for the future

because I [ __ ] hate it, but it does beg the question of whether

you trained AI on a specific person's

like language usage if then the AI would adopt that voice.

No. Like if you like if you if you don't give it access to anyone

else's use of language but one person. Okay. Would it talk like them?

Cuz like one of the one of the experiments that I was running is I have

a fairly hefty body of bad writing that I have collected over the years that I

copied and pasted into a document and said this is your style guide. Okay. like take the way that I write.

Yeah. Dissect it, understand it, and copy it to the best of your ability

and it wasn't able to do it. It does parts of it. Like there there are a lot of times where I will look at

a sentence and be like, "Yeah, that's a sentence that I would have written because I'm cringing at it right now." Yeah. Um but then also it just like forgets that.

It doesn't forget it. It's still mathematically correct. Okay. Right. It's just that they sprinkle in

little bits of stuff to make it sound a little bit more like me, right? Okay, that makes Right. And so

like the paragraphs are still the same size and the sentences all have kind of the same cadence

and like it's kind of that thing where like our old manager will post a response in chat that is very clearly

designed by chat GPT and both of us [ __ ] groan because it is so obvious

immediately that this was an AI generated. Oh in a sentence in a second in a second it's just like oh that hate

in a second. Yeah we we've even formatting I'm like no. Yeah,

it's the formatting that gives it away every time. If he took out like the the page breaks in the bold, I'd be like,

"Oh, okay. Maybe it was Oh, yeah. Like the GPT aligned rule or whatever." Yeah. Yeah. Yeah.

Like half of it. Yeah. I was going to say it's it's like it's like the AI will still write something

in an AI style, but it'll just change its adjectives and verbs to be more what you would use.

Yep. Yeah. Okay. miserable. So yeah, you'd have to remove all of its knowledge of language and just be like

this is the only person in the world who speaks English. This is how English is. Make words like them and then it might do it.

Well, then it doesn't have enough data then to like you would need to have somebody

with like a greater library than Shakespeare's

to be able to put it together. I think Shakespeare would actually be a terrible example because he is a person

who rather consistently made up his own words and restructured grammar as he saw fit.

I was going to say he also stole a lot of his work, too. So, I'm trying to think of like a Oh, okay. I

I hate this example, but if you just uploaded all of Brandon Sanderson's books. Now, the problem is

Brandon Sanderson writes in a very Brandon Sanderson already sounds like

AI. Yeah, I was gonna say he writes in such a weirdly generic way that like it actually wouldn't help

because yeah, his like I don't listen, I'm not knocking

the guy for being a prolific writer. I'm just saying that in order to produce as much content as he does, you cannot

waste a lot of time on pros. Yeah. you know, he he is the opposite of your George RR

Martins. He's the Yeah, he's the opposite of your your you know,

JR Tolkens. He's Yes, he's creating worlds, but they are worlds that are

described in a flash and not in depth. Well, that made him a really interesting

choice to finish The Wheel of Time as well because Robert Jordan writes in such a very specific way. Yeah,

that the the tonal whiplash of getting into Sanderson's work took a minute for

a lot of fans. Yeah. Because you you had to go from like an incredibly overroought, prosy,

super descriptive, very very slow, meandering description

style. Yeah. To something that sounds like he's describing an anime.

Mhm. He's He's telling you about how somebody's running on a wall and then punches somebody to death six times.

Like it's Yeah. Like Yeah. I I love a lot of Brandon Sanderson's work. I do not love his pros. And I

think that he would be a really poor choice to try and train. You know what? Maybe he's the problem. Maybe he's what

they used. That actually that that makes the most sense of anything we've said tonight is that like, oh, the problem that we have

with Chach GBD is that it writes like a guy who doesn't write particularly well.

Yeah. Yeah. And I think that like one of one of the things that I I really

have to pull back on with like the use of AI in terms of role playing games in

specific is that like efficiency is something that we can

chase to a point. Sure. But ultimately

DMing is a human activity, right? It's something that you are bringing

your own experience and thought and mastery of to the table. And

so like sometimes using a tool to make that go a little faster seems like a

good idea. And sometimes it's a a fun shortcut and sometimes you're able to like crib things from other people and

steal stuff, but like it isn't meant to be efficient,

right? Yeah. Right. There's a portion of the lonely fun

Yeah. of DMing where you're just like sitting at home doodling in a notebook like little stat blocks and like fiddling

with the math a little bit and trying to make it like more challenging in this way and less challenging in this way, right? Like that act

is artistic and deeply human. Yeah, I I completely agree. That's been my assertion about like DMing in general

when it comes to like the usage of AI is that DMing is an improvisational art

form. Um, and it is meant to be expressive in the same way that creating

and embodying a character is for a player. Um, you'll often, you know, there's

definitely different levels to it where people are engaging with the world building from just like, hey, this is a

pre-generated adventure that I'm going to run for my friends to, hey, I have my own, you know, Obsidian portal that I've

spent a decade building the lore of this completely constructed universe in. But

all all along that spectrum there is um

engagement and and artistic license taken and

your DMing style comes out that gets

just sort of cannibalized when you when you shortcut in this way. Um and this is

coming from me who freely steals information and ideas, right? Like

I still have to take them and make them my own. Yeah. Okay. I I think that there's like

Okay. There's a lot of meat on this topic that I would like love to get into. We might need to do a part two on this.

I think so. I mean, we didn't even get halfway through your notes on this, but yeah, we're in for another part two

because there's a lot of [ __ ] to talk about AI. So, we haven't even gotten into like

inshitification yet. Yeah, we we briefly touched on the idea that they are [ __ ] back in the pool

that they drink from. But yeah. All right. Well, let's grab a [ __ ] straw,

I guess. Hey, thanks for making it all the way

through this episode of Noplotonly Lore. If you're looking for more, you can always find us at noplotonlore.com

and on all the very best podcast platforms. If you like what you heard today, please

share, rate, and review the show to feed my never- ending need for attention and validation.