I was just wondering which points in their terms you found noteworthy. I've learned some stuff from others here, things that had not occurred to me, different types of usage. You seem to have a particularly negative view of it, which is fine - but I'd be interested in your elaboration, especially versus drip-feeding vague comments - you know, actual discussion.
Regarding your social engineering point - I don't see that as more than flavour of the month marketing, I don't think many modern companies really do - it's a marketing angle because they're selling things.
Hell, this place "officially" has an inclusive slant (which I AM in favour of). Do I think Wotc actually mean it? Not at all, racism keeps on sneaking through into the products.
Anyhoo, I don't understand you entering the conversation to not elaborate your views - you point to the TOS, but then shut down, what treasure/trap/mimic have you found?
I remember the first public release where it claimed to be an "algorithmic author" that took keywords from incoming text and matched patterns of responses in its pattern sets and relying on direct training data access as well. That last part caused it to regurgitate a lot of existing material.
One of its recent updates removed that link (or so it claims but it also still claims that's just what its developers told it and it has no self-awareness to be able to verify the claim). So instead (if true), we now see just the pattern sets that were built from the training data and not clips of the original data mashed together like before.
...but it still matches keyword patterns from incoming text to response patterns in its pattern sets.
Yet as shown in this thread, that can be useful, but also as shown in this thread, it shouldn't be the only reliance when DMing. I'm of the opinion that the less it's used, the better: Something to fill a gap in a pinch with its ability to match patterns but not something to do 30%+ of the work for us.
I keep harping on nuance. Humans know what nuance is and use it all the time, but we've yet to be able to give it a definition that can be turned into algorithms. Nuance adds chaos to patterns. It also makes stories and all forms of art interesting. In small quantities, the absence of nuance will go unnoticed, and that's where natural language algorithms can shine. The key phrase there is "in small quantities".
Trivia: You can supply keywords without complete sentences (such as one might type for a Google search), and it'll still give a response as if you typed in perfect grammar.
Rollback Post to RevisionRollBack
Human. Male. Possibly. Don't be a divider. My characters' backgrounds are written like instruction manuals rather than stories. My opinion and preferences don't mean you're wrong. I am 99.7603% convinced that the digital dice are messing with me. I roll high when nobody's looking and low when anyone else can see.🎲 “It's a bit early to be thinking about an epitaph. No?” will be my epitaph.
Interesting, it's not "my field" per-se, but on the subject of nuance - especially when creating images (NOT art, but images - there's no way it's "art"), I like abstract keywords as a starting point just to see what it spits out, then trying to refine an image from there. It often does "pretty good".
Excuse my.. the prompts are all there to see + with the discussion here, it kinda spurred me to generate some text with it - again the text prompt is clear to see and the poem it spat out, well - I'm pretty impressed.
If I were to use that in-game, I'd stick it into GIMP "age" the picture + stick it into a book/page/handout + voila enjoy the mysterious page you found down the side of your bed in the Tavern. It's things like that I really like it for - I'd never be able to do the image, just not my skill set.
Also, imagine "the beauty" of "make me the Monster Manual in the style of..." and it spits it out with a consistent look of your favourite artist(s). Me I'd go "goth" gimme a Gustave Dore/Albrecht Durer Monster manual please, or a Dave McKean style + you know what, it's 2023, we can do print on demand. Hypothetically at least, it'd be nice to see Wotc run with something like this and licence artist names for product use.
Interesting, it's not "my field" per-se, but on the subject of nuance - especially when creating images (NOT art, but images - there's no way it's "art"), I like abstract keywords as a starting point just to see what it spits out, then trying to refine an image from there. It often does "pretty good".
Excuse my.. the prompts are all there to see + with the discussion here, it kinda spurred me to generate some text with it - again the text prompt is clear to see and the poem it spat out, well - I'm pretty impressed.
If I were to use that in-game, I'd stick it into GIMP "age" the picture + stick it into a book/page/handout + voila enjoy the mysterious page you found down the side of your bed in the Tavern. It's things like that I really like it for - I'd never be able to do the image, just not my skill set.
Also, imagine "the beauty" of "make me the Monster Manual in the style of..." and it spits it out with a consistent look of your favourite artist(s). Me I'd go "goth" gimme a Gustave Dore/Albrecht Durer Monster manual please, or a Dave McKean style + you know what, it's 2023, we can do print on demand. Hypothetically at least, it'd be nice to see Wotc run with something like this and licence artist names for product use.
Back on topic...
Because we don't think in terms of massive pattern sets, something that "thinks" by matching massive numbers of pattern sets can pop out things we might not expect. So in that aspect, it can be good, but once the patterns are seen (which is done by repetition), it becomes obvious that it's not "thinking" at all, and its use as a creative tool on the topic ceases (though in my mind, it wasn't really being creative as it never was really pliant as DMs need to be).
Its use as a quick patch job is still there, though.
Now, licensing artwork... that's a subject that legal systems never fully managed to handle before the mashup algorithms gained a foothold (as far as I've yet encountered, so maybe someone out there actually got it right and nobody's talked about it... which is difficult to fathom but who knows...). Now, it might be too late to try to play catch up.
The lines between pattern sets and pulling directly from training data is complicated and makes my head spin just thinking about it (or it's because of the allergy season). My current understanding is that pattern sets do not equate to using sourced data as they only create reference points in a massive cobweb between words. I'm inclined to agree with that pattern sets aren't infringing copyrights... for the moment.
I don't know how "AI" art is able to replicate imagery if it cannot access sourced data. The massive webwork of pattern sets just for a 1-dimensional prompt is obscenely complicated and it simply returns a 1-dimensional response of the same data type. To cross from a 1-dimensional data type of word references to a 2-dimensional data type of references between pixels doesn't seem possible with our current technology... unless the 2-D references are image parts and not per-pixel referencing... and that's a legal minefield.
Human. Male. Possibly. Don't be a divider. My characters' backgrounds are written like instruction manuals rather than stories. My opinion and preferences don't mean you're wrong. I am 99.7603% convinced that the digital dice are messing with me. I roll high when nobody's looking and low when anyone else can see.🎲 “It's a bit early to be thinking about an epitaph. No?” will be my epitaph.
Definitely back on to topic - and apologies for my part in that.
It's late here + math is not my strong point at all, but as a digression AND it seems you understand this all far better than me. I'm just the fanboy cheering from the sidelines.
Definitely back on to topic - and apologies for my part in that.
It's late here + math is not my strong point at all, but as a digression AND it seems you understand this all far better than me. I'm just the fanboy cheering from the sidelines.
What happens if/when? Can I live in a Star Trek/holodeck D&D paradise?
I read that. Such a limited view. They need to give that technology over to people who don't know computing. All that particular group wants is a faster computer, so blindingly fast that it could calculate a million times more possibilities than current modern computers.
That doesn't solve the problem of bridging absolute digital processing into a relative analog world. People want to go with what they understand, and us techies get locked into old ways.
We need people who understand things differently... or possibly completely don't understand things the way current "experts" do and the "experts" need to listen instead of dismiss people who suggest trying something the "experts" don't understand.
If a model doesn't provide the right results, adjust it a few times, but if it continues to come up short, you need a new model... or you'll end up with a fictitious planet closer to the Sun than Mercury because they kept adjusting the models instead of questioning it... with the answer being something quite simple as to why Mercury was doing what it did... a completely different model that actually fit. (The vast majority of people abandoned belief in Vulcan by the turn of the last century, but I had an instructor that still swore by it because it was published in NY Post that "its existence could no longer be denied" half a century prior.)
Our models are all still based on one thing: Binary. After all this time all the way back to the earliest devices that can be considered digital, the binary model has fallen short... but people keep insisting that it just needs to be properly adjusted to make it work.
Digital devices have consistently shown they can't go any further than just getting faster. Why won't we try something new? Quantum "bits" aren't just 2-state, but the greater research community has decided that's not useful... because it can't be used digitally, ignoring that maybe the digital part is the problem and not the variable state of the bits. People were able to create compute nodes with living neurons... but what did they do with it? Try to make it work like digital devices, and it was so incredibly inefficient that the concept is all but abandoned now. Such a waste. Living neurons work properly by relative frequencies triggering more frequencies when inconsistent and seemingly arbitrary boundaries are crossed, not binary like the sci-fi stories say.
With a "living" compute node on the node's own terms instead of squeezing it into a binary box, "unhandled exceptions" should become a thing of the past. Life has no unhandled exceptions. Binary code does all the time and does one of three things with it: Try to ignore it, try to walk back the process and give it another go, or throw an error and give up... all dependent on what it was told to do when a situation that wasn't expected is encountered.
ChatGPT attempts to simulate through a whole process of millions of calculations for what we do without a whole process of millions of calculations... and like all digital processes for ages (even before electric computing), it still falls short. The binary model is showing us that it just doesn't work like that, but people are loathe to think of anything else even though it's standing right in front of us (if we get off of our computers and phones).
It'll forever be a tool, not a companion, not a think-tank, not an artist, not a storyteller.
Rollback Post to RevisionRollBack
Human. Male. Possibly. Don't be a divider. My characters' backgrounds are written like instruction manuals rather than stories. My opinion and preferences don't mean you're wrong. I am 99.7603% convinced that the digital dice are messing with me. I roll high when nobody's looking and low when anyone else can see.🎲 “It's a bit early to be thinking about an epitaph. No?” will be my epitaph.
Definitely back on to topic - and apologies for my part in that.
It's late here + math is not my strong point at all, but as a digression AND it seems you understand this all far better than me. I'm just the fanboy cheering from the sidelines.
What happens if/when? Can I live in a Star Trek/holodeck D&D paradise?
I read that. Such a limited view. They need to give that technology over to people who don't know computing. All that particular group wants is a faster computer, so blindingly fast that it could calculate a million times more possibilities than current modern computers.
That doesn't solve the problem of bridging absolute digital processing into a relative analog world. People want to go with what they understand, and us techies get locked into old ways.
We need people who understand things differently... or possibly completely don't understand things the way current "experts" do and the "experts" need to listen instead of dismiss people who suggest trying something the "experts" don't understand.
If a model doesn't provide the right results, adjust it a few times, but if it continues to come up short, you need a new model... or you'll end up with a fictitious planet closer to the Sun than Mercury because they kept adjusting the models instead of questioning it... with the answer being something quite simple as to why Mercury was doing what it did... a completely different model that actually fit. (The vast majority of people abandoned belief in Vulcan by the turn of the last century, but I had an instructor that still swore by it because it was published in NY Post that "its existence could no longer be denied" half a century prior.)
Our models are all still based on one thing: Binary. After all this time all the way back to the earliest devices that can be considered digital, the binary model has fallen short... but people keep insisting that it just needs to be properly adjusted to make it work.
Digital devices have consistently shown they can't go any further than just getting faster. Why won't we try something new? Quantum "bits" aren't just 2-state, but the greater research community has decided that's not useful... because it can't be used digitally, ignoring that maybe the digital part is the problem and not the variable state of the bits. People were able to create compute nodes with living neurons... but what did they do with it? Try to make it work like digital devices, and it was so incredibly inefficient that the concept is all but abandoned now. Such a waste. Living neurons work properly by relative frequencies triggering more frequencies when inconsistent and seemingly arbitrary boundaries are crossed, not binary like the sci-fi stories say.
With a "living" compute node on the node's own terms instead of squeezing it into a binary box, "unhandled exceptions" should become a thing of the past. Life has no unhandled exceptions. Binary code does all the time and does one of three things with it: Try to ignore it, try to walk back the process and give it another go, or throw an error and give up... all dependent on what it was told to do when a situation that wasn't expected is encountered.
ChatGPT attempts to simulate through a whole process of millions of calculations for what we do without a whole process of millions of calculations... and like all digital processes for ages (even before electric computing), it still falls short. The binary model is showing us that it just doesn't work like that, but people are loathe to think of anything else even though it's standing right in front of us (if we get off of our computers and phones).
It'll forever be a tool, not a companion, not a think-tank, not an artist, not a storyteller.
First, I agree.
Second as a tool things become a little muddy. There are plenty of relevant examples, my personal favorite is ~If the only tool you have is a hammer every problem is a nail; unless you choose to look for a better tool(solution).
To further muddy waters, there are plenty of "tools" that are only for sale because of profit potential everything else is tertiary.
Rollback Post to RevisionRollBack
CENSORSHIP IS THE TOOL OF COWARDS and WANNA BE TYRANTS.
My party perform a particularly heroic feat, destroying the Ice Temple, a month or so later they're in Bryn Shander when they hear this
(Verse 1) In a land shrouded by frost and snow, Where the bitter winds forever blow, An ancient temple of ice stands tall, Guarding secrets, ready to enthrall.
(Pre-Chorus) But in the darkness, a light will shine, A band of heroes, brave and divine, With hearts of fire, they'll face the storm, To break the ice and reform the norm.
(Chorus) Heroes rising, their spirits high, Through the blizzard, they'll touch the sky, With courage blazing, they'll break the mold, The ice temple's fate, they will unfold.
(Verse 2) Legends whispered, tales of the past, Of warriors bold who'd faced the blast, They faced the temple's icy wrath, In search of power, a righteous path.
(Pre-Chorus) They ventured forth, through the frozen gate, Guided by destiny, fueled by fate, Together united, they forged ahead, Ready to face the cold and dread.
(Chorus) Heroes rising, their spirits high, Through the blizzard, they'll touch the sky, With courage blazing, they'll break the mold, The ice temple's fate, they will unfold.
(Bridge) In the temple's heart, a diamond gleamed, Its power bound, a force esteemed, But the heroes knew, it must be freed, To heal the land, fulfill the need.
(Verse 3) Through icy chambers, they fought and strived, Braving trials, determined to survive, Each hero using their unique skill, Their purpose unwavering, an iron will.
(Pre-Chorus) They faced the guardian, a mighty foe, A creature ancient, with an icy glow, With strength combined, they stood as one, Their battle cry echoed, the ice undone.
(Chorus)
*ChatGPT write a song about heroes destroying an ice temple. The party don't know it's AI, if they're "on point" with their RP - the characters are going to be pleased/flattered + the actual players should think it's pretty cool too. I would tell them (after) it was AI, but if all else is equal... It's the suspension of disbelief thing - can you believe an Evil Ice Goddess, atop a vengeful Dragon, breathing icy death. Now, that's just a nice little handout - I'm not singing/recording etc. etc. songwriting's a stretch for me and that would have taken me maybe an hour or so to "get happy" with it.
I've tested it moderately and it can more/less nail Lovecraft, Harold Pinter or Michael Bay (Bay script was pretty funny too).
EDIT TO ADD: write a good-natured trans agenda in the style of H.P. Lovecraft Came out pretty well too, genuinely seems good-natured (humanist at least) and seems to take in some of the Dreamlands writing as an influence, but that may just be my "wanting" to see it.
My experience is limited and based off GPT4 usage. I was pretty impressed with the 'possibilities' that are now available. As with everything, there will be millions of attempted implementations, each with their own pros/cons. One thing is for sure, it's going to happen. The question is when?
Things I tried and reviewed.
Map creation both true and '3D' top-down views. World, Kingdom, City, and Dungeon themed. I used keywords to create maps for a specific race and mixture of races. It actually was impressive in this aspect. Names of river, cities, lakes, ocean where correctly generated for the race that occupied that area of the map. Things were generally well proportioned on the map. You could even include a legend with map highlights, compass, etc. Dungeon based images were equally impressive - lighting and shadows - centralized goblin camp in large cavern and a couple outposts in smaller nearby caverns.
2D Image creation - The amount of detail rendered given input based on race, class, hair/eye color, size, weapon, armor was pretty impressive.
3D Image creation - The least lackluster of them all, but of course the hardest to achieve. It wrote a script for Blender that I was able to run and import my first dwarf with a hammer. Very basic input, I ran out of GPT usages for a bit. lol. The first attempt rendered two blobs for the head/neck and body. And a very basic hammer by their side. Not much of a start, but it is where you'd start!
So many possibilities. Overall - the image/map creation can save a ton of time. If you feed GPT a basic storyline, it can expand it pretty well. Not perfect but I can see where this could help shorten the time taken to create it all yourself. I can't really speak much to the ability to create 3D models as I only had one try before running out of GPT requests for the moment. I'm guessing I might get a little better result with refined attempts, but we'll see!
I should add that I am a novice (at best) in using both GPT and Blender.
Rollback Post to RevisionRollBack
To post a comment, please login or register a new account.
Yeah, have been - will do etc.
I was just wondering which points in their terms you found noteworthy. I've learned some stuff from others here, things that had not occurred to me, different types of usage. You seem to have a particularly negative view of it, which is fine - but I'd be interested in your elaboration, especially versus drip-feeding vague comments - you know, actual discussion.
Regarding your social engineering point - I don't see that as more than flavour of the month marketing, I don't think many modern companies really do - it's a marketing angle because they're selling things.
Hell, this place "officially" has an inclusive slant (which I AM in favour of). Do I think Wotc actually mean it? Not at all, racism keeps on sneaking through into the products.
Anyhoo, I don't understand you entering the conversation to not elaborate your views - you point to the TOS, but then shut down, what treasure/trap/mimic have you found?
https://wulfgold.substack.com
Blog - nerd stuff
https://deepdreamgenerator.com/u/wulfgold
A.I. art - also nerd stuff - a gallery of NPC portraits - help yourself.
I remember the first public release where it claimed to be an "algorithmic author" that took keywords from incoming text and matched patterns of responses in its pattern sets and relying on direct training data access as well. That last part caused it to regurgitate a lot of existing material.
One of its recent updates removed that link (or so it claims but it also still claims that's just what its developers told it and it has no self-awareness to be able to verify the claim). So instead (if true), we now see just the pattern sets that were built from the training data and not clips of the original data mashed together like before.
...but it still matches keyword patterns from incoming text to response patterns in its pattern sets.
Yet as shown in this thread, that can be useful, but also as shown in this thread, it shouldn't be the only reliance when DMing. I'm of the opinion that the less it's used, the better: Something to fill a gap in a pinch with its ability to match patterns but not something to do 30%+ of the work for us.
I keep harping on nuance. Humans know what nuance is and use it all the time, but we've yet to be able to give it a definition that can be turned into algorithms. Nuance adds chaos to patterns. It also makes stories and all forms of art interesting. In small quantities, the absence of nuance will go unnoticed, and that's where natural language algorithms can shine. The key phrase there is "in small quantities".
Trivia: You can supply keywords without complete sentences (such as one might type for a Google search), and it'll still give a response as if you typed in perfect grammar.
Human. Male. Possibly. Don't be a divider.
My characters' backgrounds are written like instruction manuals rather than stories. My opinion and preferences don't mean you're wrong.
I am 99.7603% convinced that the digital dice are messing with me. I roll high when nobody's looking and low when anyone else can see.🎲
“It's a bit early to be thinking about an epitaph. No?” will be my epitaph.
Interesting, it's not "my field" per-se, but on the subject of nuance - especially when creating images (NOT art, but images - there's no way it's "art"), I like abstract keywords as a starting point just to see what it spits out, then trying to refine an image from there. It often does "pretty good".
Excuse my.. the prompts are all there to see + with the discussion here, it kinda spurred me to generate some text with it - again the text prompt is clear to see and the poem it spat out, well - I'm pretty impressed.
https://deepdreamgenerator.com/ddream/gubf64mhod9
If I were to use that in-game, I'd stick it into GIMP "age" the picture + stick it into a book/page/handout + voila enjoy the mysterious page you found down the side of your bed in the Tavern. It's things like that I really like it for - I'd never be able to do the image, just not my skill set.
Also, imagine "the beauty" of "make me the Monster Manual in the style of..." and it spits it out with a consistent look of your favourite artist(s). Me I'd go "goth" gimme a Gustave Dore/Albrecht Durer Monster manual please, or a Dave McKean style + you know what, it's 2023, we can do print on demand. Hypothetically at least, it'd be nice to see Wotc run with something like this and licence artist names for product use.
https://wulfgold.substack.com
Blog - nerd stuff
https://deepdreamgenerator.com/u/wulfgold
A.I. art - also nerd stuff - a gallery of NPC portraits - help yourself.
Back on topic...
Because we don't think in terms of massive pattern sets, something that "thinks" by matching massive numbers of pattern sets can pop out things we might not expect. So in that aspect, it can be good, but once the patterns are seen (which is done by repetition), it becomes obvious that it's not "thinking" at all, and its use as a creative tool on the topic ceases (though in my mind, it wasn't really being creative as it never was really pliant as DMs need to be).
Its use as a quick patch job is still there, though.
Now, licensing artwork... that's a subject that legal systems never fully managed to handle before the mashup algorithms gained a foothold (as far as I've yet encountered, so maybe someone out there actually got it right and nobody's talked about it... which is difficult to fathom but who knows...). Now, it might be too late to try to play catch up.
The lines between pattern sets and pulling directly from training data is complicated and makes my head spin just thinking about it (or it's because of the allergy season). My current understanding is that pattern sets do not equate to using sourced data as they only create reference points in a massive cobweb between words. I'm inclined to agree with that pattern sets aren't infringing copyrights... for the moment.
I don't know how "AI" art is able to replicate imagery if it cannot access sourced data. The massive webwork of pattern sets just for a 1-dimensional prompt is obscenely complicated and it simply returns a 1-dimensional response of the same data type. To cross from a 1-dimensional data type of word references to a 2-dimensional data type of references between pixels doesn't seem possible with our current technology... unless the 2-D references are image parts and not per-pixel referencing... and that's a legal minefield.
Human. Male. Possibly. Don't be a divider.
My characters' backgrounds are written like instruction manuals rather than stories. My opinion and preferences don't mean you're wrong.
I am 99.7603% convinced that the digital dice are messing with me. I roll high when nobody's looking and low when anyone else can see.🎲
“It's a bit early to be thinking about an epitaph. No?” will be my epitaph.
Definitely back on to topic - and apologies for my part in that.
It's late here + math is not my strong point at all, but as a digression AND it seems you understand this all far better than me. I'm just the fanboy cheering from the sidelines.
https://www.bbc.co.uk/news/science-environment-64492456
What happens if/when? Can I live in a Star Trek/holodeck D&D paradise?
https://wulfgold.substack.com
Blog - nerd stuff
https://deepdreamgenerator.com/u/wulfgold
A.I. art - also nerd stuff - a gallery of NPC portraits - help yourself.
I read that. Such a limited view. They need to give that technology over to people who don't know computing. All that particular group wants is a faster computer, so blindingly fast that it could calculate a million times more possibilities than current modern computers.
That doesn't solve the problem of bridging absolute digital processing into a relative analog world. People want to go with what they understand, and us techies get locked into old ways.
We need people who understand things differently... or possibly completely don't understand things the way current "experts" do and the "experts" need to listen instead of dismiss people who suggest trying something the "experts" don't understand.
If a model doesn't provide the right results, adjust it a few times, but if it continues to come up short, you need a new model... or you'll end up with a fictitious planet closer to the Sun than Mercury because they kept adjusting the models instead of questioning it... with the answer being something quite simple as to why Mercury was doing what it did... a completely different model that actually fit. (The vast majority of people abandoned belief in Vulcan by the turn of the last century, but I had an instructor that still swore by it because it was published in NY Post that "its existence could no longer be denied" half a century prior.)
Our models are all still based on one thing: Binary. After all this time all the way back to the earliest devices that can be considered digital, the binary model has fallen short... but people keep insisting that it just needs to be properly adjusted to make it work.
Digital devices have consistently shown they can't go any further than just getting faster. Why won't we try something new? Quantum "bits" aren't just 2-state, but the greater research community has decided that's not useful... because it can't be used digitally, ignoring that maybe the digital part is the problem and not the variable state of the bits. People were able to create compute nodes with living neurons... but what did they do with it? Try to make it work like digital devices, and it was so incredibly inefficient that the concept is all but abandoned now. Such a waste. Living neurons work properly by relative frequencies triggering more frequencies when inconsistent and seemingly arbitrary boundaries are crossed, not binary like the sci-fi stories say.
With a "living" compute node on the node's own terms instead of squeezing it into a binary box, "unhandled exceptions" should become a thing of the past. Life has no unhandled exceptions. Binary code does all the time and does one of three things with it: Try to ignore it, try to walk back the process and give it another go, or throw an error and give up... all dependent on what it was told to do when a situation that wasn't expected is encountered.
ChatGPT attempts to simulate through a whole process of millions of calculations for what we do without a whole process of millions of calculations... and like all digital processes for ages (even before electric computing), it still falls short. The binary model is showing us that it just doesn't work like that, but people are loathe to think of anything else even though it's standing right in front of us (if we get off of our computers and phones).
It'll forever be a tool, not a companion, not a think-tank, not an artist, not a storyteller.
Human. Male. Possibly. Don't be a divider.
My characters' backgrounds are written like instruction manuals rather than stories. My opinion and preferences don't mean you're wrong.
I am 99.7603% convinced that the digital dice are messing with me. I roll high when nobody's looking and low when anyone else can see.🎲
“It's a bit early to be thinking about an epitaph. No?” will be my epitaph.
First, I agree.
Second as a tool things become a little muddy. There are plenty of relevant examples, my personal favorite is ~If the only tool you have is a hammer every problem is a nail; unless you choose to look for a better tool(solution).
To further muddy waters, there are plenty of "tools" that are only for sale because of profit potential everything else is tertiary.
CENSORSHIP IS THE TOOL OF COWARDS and WANNA BE TYRANTS.
So, tangentially - confirmation bias.
My party perform a particularly heroic feat, destroying the Ice Temple, a month or so later they're in Bryn Shander when they hear this
(Verse 1) In a land shrouded by frost and snow, Where the bitter winds forever blow, An ancient temple of ice stands tall, Guarding secrets, ready to enthrall.
(Pre-Chorus) But in the darkness, a light will shine, A band of heroes, brave and divine, With hearts of fire, they'll face the storm, To break the ice and reform the norm.
(Chorus) Heroes rising, their spirits high, Through the blizzard, they'll touch the sky, With courage blazing, they'll break the mold, The ice temple's fate, they will unfold.
(Verse 2) Legends whispered, tales of the past, Of warriors bold who'd faced the blast, They faced the temple's icy wrath, In search of power, a righteous path.
(Pre-Chorus) They ventured forth, through the frozen gate, Guided by destiny, fueled by fate, Together united, they forged ahead, Ready to face the cold and dread.
(Chorus) Heroes rising, their spirits high, Through the blizzard, they'll touch the sky, With courage blazing, they'll break the mold, The ice temple's fate, they will unfold.
(Bridge) In the temple's heart, a diamond gleamed, Its power bound, a force esteemed, But the heroes knew, it must be freed, To heal the land, fulfill the need.
(Verse 3) Through icy chambers, they fought and strived, Braving trials, determined to survive, Each hero using their unique skill, Their purpose unwavering, an iron will.
(Pre-Chorus) They faced the guardian, a mighty foe, A creature ancient, with an icy glow, With strength combined, they stood as one, Their battle cry echoed, the ice undone.
(Chorus)
*ChatGPT write a song about heroes destroying an ice temple. The party don't know it's AI, if they're "on point" with their RP - the characters are going to be pleased/flattered + the actual players should think it's pretty cool too. I would tell them (after) it was AI, but if all else is equal... It's the suspension of disbelief thing - can you believe an Evil Ice Goddess, atop a vengeful Dragon, breathing icy death. Now, that's just a nice little handout - I'm not singing/recording etc. etc. songwriting's a stretch for me and that would have taken me maybe an hour or so to "get happy" with it.
I've tested it moderately and it can more/less nail Lovecraft, Harold Pinter or Michael Bay (Bay script was pretty funny too).
EDIT TO ADD: write a good-natured trans agenda in the style of H.P. Lovecraft
Came out pretty well too, genuinely seems good-natured (humanist at least) and seems to take in some of the Dreamlands writing as an influence, but that may just be my "wanting" to see it.
https://wulfgold.substack.com
Blog - nerd stuff
https://deepdreamgenerator.com/u/wulfgold
A.I. art - also nerd stuff - a gallery of NPC portraits - help yourself.
My experience is limited and based off GPT4 usage. I was pretty impressed with the 'possibilities' that are now available. As with everything, there will be millions of attempted implementations, each with their own pros/cons. One thing is for sure, it's going to happen. The question is when?
Things I tried and reviewed.
Map creation both true and '3D' top-down views. World, Kingdom, City, and Dungeon themed. I used keywords to create maps for a specific race and mixture of races. It actually was impressive in this aspect. Names of river, cities, lakes, ocean where correctly generated for the race that occupied that area of the map. Things were generally well proportioned on the map. You could even include a legend with map highlights, compass, etc. Dungeon based images were equally impressive - lighting and shadows - centralized goblin camp in large cavern and a couple outposts in smaller nearby caverns.
2D Image creation - The amount of detail rendered given input based on race, class, hair/eye color, size, weapon, armor was pretty impressive.
3D Image creation - The least lackluster of them all, but of course the hardest to achieve. It wrote a script for Blender that I was able to run and import my first dwarf with a hammer. Very basic input, I ran out of GPT usages for a bit. lol. The first attempt rendered two blobs for the head/neck and body. And a very basic hammer by their side. Not much of a start, but it is where you'd start!
So many possibilities. Overall - the image/map creation can save a ton of time. If you feed GPT a basic storyline, it can expand it pretty well. Not perfect but I can see where this could help shorten the time taken to create it all yourself. I can't really speak much to the ability to create 3D models as I only had one try before running out of GPT requests for the moment. I'm guessing I might get a little better result with refined attempts, but we'll see!
I should add that I am a novice (at best) in using both GPT and Blender.