NovelAI - AI becomes gamemaster in text based RPG. And handles everything.

[fork]

Since civilisation as we know it ends in nuclear hellfire [...]

It doesn't. It ends in globalist totalitarianism.

 

[Perkel]

Here are some tips to how to actually use it properly. If you just click "next" Without doing anything then story will go anywhere and could be anything and depending on chosen profile of AI it can lead to weird places.
What you need in order for it to be great are 3 things:

author's note
memory
lorebook

They are on right side. I talked to one of dudes who is really good at this and this is essentially what he said. It is worth noting that tutorial for novelAI is outdated and it was made for older AI that worked a bit different so what that dude said is up to date stuff.

memory - is place where you put general themes for creation. So kind of like summary of book from back cover. Thanks to this things will stay on course. If you for example write there that there no magic in this world then AI will try to avoid magic.

author's note - is essentially a write pad with strong influence on what is happening. In this place you write some very short notices if you feel that your story doesn't go into direction you want. So you can write something like "Story takes place in palace" and AI will try to gravitate toward that like you are in cave and then it will find a way for story to go to palace. Most of the time this windows stays empty because it has too big influence which can lead to story non stop focusing on what was written there.

lorebook - the real meat of what makes novelAI work great. Basically it is important memory where you can add places, people, items, events, kindoms, races and pretty much every concept you can imagine. Those are even sharable so if you want to say play your game in world of Dragon Age from bioware you can download lorebook from novelAI discord and then your game will use that world.


Lorebook being something that can be imported and exported is fucking great thing. Basically you can just download some game lorebook and just play in that world.


If you don't use any of those then AI will just default to his knowledge and what he thinks is dark elf from general internet knowledge. With some prompts you can make them as unique as you want but lorebook has advantage of being much better at referencing knowledge than prompt which can lose importance over time and at some point it can default to something else

 

[Perkel]

So for an example:

Memory:

The kingdom of Codexia, after long wars with dark elves peaceful times arrived. You are named Jack and you are knight in service of Duke Kruno. Duke Kruno was obnoxious tyrannical idiot who somehow due to his family interactions became Duke.

While on the surface everything was peaceful, deep underground dark elves planned their invasion in silence.

Author's Note:

EMPTY

Lorebook:
You can add here stuff to flesh out world more, For example if you think that Duke Kruno being idiot was not enough you add lorebook entry about him where you describe he fucked all of his sisters and his mother, he had lover who was a goat and other stuff.

 

[Dexter]

I don't really care much for text-based AI, but I stumbled upon these Screenshots of Weebs making and abusing AI characters on some I believe Google-affiliated service called Character.AI: https://www.character.ai/

Spoiler

They did this and probably err... other stuff so much that the company apparently decided to introduce Human Moderators to jump in for when things get out of hand as was posted in another thread: https://rpgcodex.net/forums/threads...ke-oil-for-morons.144412/page-22#post-8163841

Make of it what you will.

 

[Haba]

Imagine being the moderator who has to see... uh, that kind of fantasies play out.



Good thing a lot of people from Facebook are looking for new career opportunities.

 

[Maxwell's Demon]

Imagine being the moderator who has to see... uh, that kind of fantasies play out.

Isn't half the reason that people become internet moderators because they want a pretext to see that kind of content? (The other half is wanting a pretext to inflict suffering on others for fun.)

 

[Sarathiour]

Example articles I've had the AI write:

Elon Musk Says Bridge In London For Sale By 2023

Elon Musk Says Bridge In London For Sale By 2023

"In just one year, you'll all be able to buy your own bridge in London."

You get a bridge, and you get a bridge, and you get a bridge.

That's the deal with Elon Musk's Boring Company, which is building a new underground transit system for Los Angeles that will include a high-speed tunnel network connecting major cities like LAX, downtown, and Westwood. The company also has plans to build a Hyperloop between New York City and Washington, DC.

And now it looks like they're going international.

The Boring Company announced today that its first overseas project will be a series of several hundred bridges across the Thames River in London. The plan calls for custom-designed structures that can span the river at various points along the route, allowing pedestrians and vehicles to cross without having to go up or down any stairs or ramps.

"Each bridge will be custom-made from carbon fiber and other lightweight materials," said Musk. "We've already started working on them."

Musk says he expects the entire project to cost less than $1 billion—a fraction of what it would normally take to build an elevated road over the Thames. He added that the bridges should be ready by 2023, assuming everything goes according to schedule.

"I'm excited about this project because it's not only about creating something cool; it's about improving people's lives," said Musk. "Imagine how much time you'd save if you didn't have to walk up and down stairs every day. Or how many calories you could burn walking instead of taking the tube."

As part of his pitch, Musk showed off some concept renderings of the bridges, including one that was designed to look like a giant pair of scissors crossing the river.

"It's really hard to make a good-looking bridge," Musk admitted. "But we think these are pretty sweet."

Musk says the bridges should be completed by 2023. That's only one year from now!

"Londoners have struggled with heavy traffic for decades," he said on Twitter. "With the Boring Company tunnels, we can bring fast travel directly to them. We are excited to work with Transport for London and the Mayor of London to make this vision a reality."

The Chemical Composition of Lloth's Sweat and How It Might Smell

The Chemical Composition of Lloth's Sweat and How It Might Smell

By: Dr. J.R. "Bobby" Bannister, Ph.D., M.S., F.I.C.A.

The following is a scientific analysis of the chemical composition of the sweat produced by the goddess Lloth, as well as an attempt to describe its odor. The purpose of this paper is not to provide a definitive answer to the question of whether or not Lloth has a body odor (although I would argue that she does), but rather to provide a basis for discussion on the subject.

The first step in any scientific investigation of a problem is to define the parameters under which you are going to work. In this case we will be looking at the sweat produced by the goddess Lloth. This is important because there have been some conflicting reports about her physical appearance. Some claim that she looks like a Drow female while others say that she looks more like a spider than anything else. For our purposes here, we will assume that she looks like a Drow female.

Next we need to determine what kind of information we want to get about the sweat. We could ask questions such as: What is the color of the sweat? Is it clear or cloudy? Does it smell bad? Is it salty or sweet? Or perhaps we might want to know if it contains any toxins or poisons. There are many possible questions that we could ask, but for now let us limit ourselves to just two: What is the chemical composition of the sweat? And how does it smell?

The next step is to gather all the data that we can find on the subject. Since Lloth only exists within the pages of the Advanced Dungeons & Dragons game manuals, we will use those books as our primary source of information. Fortunately, they contain quite a bit of information about her.

Lloth is described as having long white hair, pale skin, and red eyes. Her height is given as 6'2", making her taller than most Drow females. No official measurement of her weight is given, but based on other information in the manuals, she would likely weigh around 150 to 200 pounds (68 to 91 kilograms).

Given that we don't really know much about her physiology, we will make several assumptions about her body chemistry. First, we will assume that she produces sweat through the same mechanism as humans do. Second, we will assume that her sweat is water-based.

There are three major components of sweat: water, sodium chloride, and potassium chloride. These are the main electrolytes in sweat. They are also present in blood plasma. If we were to analyze a sample of Lloth's sweat, we would expect to see these three substances present in roughly equal amounts.

The next thing we should look for is the presence of any toxins or poisons that might alter the chemical makeup of the sweat. Unfortunately, Lloth is not listed as being able to produce any known toxins or poisons in Drow form. Therefore, we can safely rule out any possibility of finding any toxins or poisons in her sweat.

Finally, we must consider the possibility that Lloth may produce some sort of pheromone. This is a substance that is secreted into the air by an animal or plant and then detected by another individual of the same species. A pheromone is often used to communicate with members of the same species. For example, when a male dog lifts his leg against a tree, he is marking his territory. Other dogs who smell this scent recognize that this particular area belongs to him and avoid it.

It's possible that Lloth might produce a pheromone that acts as a sexual attractant for other Drow. Her 22 Charisma score makes her one of the most charismatic individuals in the entire Underdark. She is certainly attractive enough to warrant the attention of any number of Drow males. However, another possibility is that she uses her pheromones to control and manipulate the behavior of spiders. As we discussed earlier, she is capable of taking the form of a giant black widow spider. Perhaps she uses her pheromones to influence the actions of spiders in general. Since there is no mention of pheromones relating to spider control in any of the D&D game books, we cannot confirm or deny this hypothesis.

Since we are unable to find any evidence of toxins or poisons, nor any pheromones, we will focus our efforts on determining the chemical composition of Lloth's sweat. To do so, we will need to examine her sweat glands.
Since the Drow are a fantasy race, we will have to rely on our knowledge of human physiology to make some reasonable guesses about Lloth's sweat glands.

The sweat glands of humans secrete a mixture of water, sodium chloride, and potassium chloride. The amount of each varies from person to person. Racially, there are few differences in the levels of these substances in the sweat of a human. For example, blacks have more chloride in sweat than do whites. However, the smell of sweat can vary greatly depending on skin color.

For example, people with darker skin tend to have larger pores than lighter skinned people. This, combined with the fact that dark skin has more apocrine glands than light skin, may explain why people with dark skin have a stronger odor than those with lighter skin. Apocrine glands excrete fat and protein along with water (Poirier et al p 567).

Based upon this, we may assume that a Drow with dark skin would have a stronger odor than an Elf with light skin.

From these observations we can assume that the average Drow would have a slightly stronger odor than an average human of Caucasian descent.

However, since we are dealing with a deity, we need to take things up a notch. Not only is Lloth an extremely powerful Drow, but she is also a goddess. Given this, we can assume that she is very attractive to other Drow, and her scent would be even more powerful than that of an average Drow.

If we combine the observations already made, we can come up with a rough estimate of Lloth's odor. She is an attractive Drow female with dark skin, who is at least 6'2" tall, and weighs about 150 to 200 pounds. Based upon these observations, we can assume that her sweat would have a higher concentration of water, sodium chloride, and potassium chloride than an average Drow.

We can further refine this assumption by estimating the size of her apocrine glands. Since she is a goddess, and tall for a member of her race, we can assume that her glands are probably somewhat larger than those of an ordinary human.
Using this information, we can come up with the following conclusions:

A Drow with dark skin would have a stronger odor than an average human.
Her sweat glands are large enough that they would produce significantly more sweat than an average human.
She has apocrine glands that are of an increased size commensurate with her above stated stature.
Due to her godly status, her body odor would be much stronger than an average Drow's.

In summary:

Her sweat would contain more water, sodium chloride, and potassium chloride than an average human. This would result in a stronger and more pronounced odor.

Based upon the observations made thus far, I believe that my final conclusion is sound: Lloth's sweat smells strongly like vinegar. A pungent aroma indeed.

New article written for this thread in a few minutes:

Why Are Certain People So Butthurt By AI?​

By David Lumb

For decades now, the long-standing concern about artificial intelligence has been that it might escape from its electronic cages and run amok. It's a fear we have on every new wave of technology — remember Skynet in "The Terminator?" And with the AI boom, the worry is even more pervasive than usual. In 2012, a Pew survey found that 45 percent of Americans believed that a superintelligent AI would be "bad" for society; by last year, that percentage had risen to 62 percent.

And now some scientists are starting to ask themselves: Why? How did certain people get so worked up about a phenomenon that, after all, still lacks many basic features of consciousness? It's a fair question. After all, concern about robots doesn't seem to have extended to the loss of human jobs over the last two centuries or the decline of unionization. Indeed, as Yale researcher Oren Etzioni pointed out earlier this year in an essay published in the Harvard Business Review, concern about the rise of robots has been growing much faster among those with less education and income. As Etzioni put it, "artificial intelligence is something only other intellectuals care about."

Why should that be? The answer is simple and depressing: Those who fear technological progress generally come from poor backgrounds, where change can mean losing what you already have rather than gaining a new opportunity. People like this tend to view technology with suspicion and anxiety — just look at conspiracy theories or anti-vaxxers on Facebook — because they believe that the world is set up to screw them over.

"Seems to generate utter gibberish to me." wrote one angry reader when I posted my column online last week. That comment was typical; most comments were negative or dismissive, without any real attempt to understand how AIs might actually work or their potential applications. This isn't surprising given the low level of understanding about AI in general. One survey found that while 80 percent of Americans believe robots will be common household items by 2025, only 17 percent could define AI, and even fewer could explain its basic capabilities.

This lack of knowledge helps fuel the irrational fears about superintelligent machines. In fact, as we've discussed before, there's no evidence that anything like true artificial intelligence is coming anytime soon (if ever). But that hasn't stopped some people from claiming it's here — including the billionaire Tesla founder Elon Musk, who recently told Rolling Stone that "we have AI already, sort of." As you may recall, his logic goes like this:

"It is a fundamental mistake to assume that current deep learning neural networks are anything other than narrow AI — i.e. AI that is qualitatively indistinguishable from human intelligence," he wrote in a blog post earlier this year.
He continued: "They can already write restaurant reviews and tell you whether one picture has a dog in it." And so, by definition, they're intelligent.

I'm sure that's comforting for Mr. Musk to hear, but if true, it would also mean that many of the world's best computer scientists and mathematicians were duped by an elaborate hoax. For what it's worth, I think they know better, which is why Musk is known more as the "Tesla-maker" than as the "AI-maker."

The real issue isn't whether machines are becoming smarter; it's how much power we give them over our lives. It's time to stop worrying about Skynet and start asking ourselves what kind of life we want to live — and then figuring out how technology will help us get there.

And this is just GPT-2. OpenAI's GPT-3 is more advanced, but proprietary.

The lolth was funny, but the conclusion of the article is the opposite of the thesis he proposed. The article is actually quite good at showing why AI got no idea of what it's talking about.

Other are just bland drivel.

 

[Non-Edgy Gamer]

The article is actually quite good at showing why AI got no idea of what it's talking about.

Well, of course it doesn't. At best, GPT will fool you that it knows what it's saying, but it's really just spitting out what it thinks is a likely continuation of your prompt.

For example, if you ask it what 1+1 is, it will probably say 2. And the same for a lot of common equations it was trained on. But if you ask it anything complex, it will likely just make up a number. It doesn't know math, it just knows a number is the response to the equation, and if there's not enough training on what that number is, it will just chose one at random or tell you it doesn't know.

 

[grimace]

DALL-E and similar software can make some amazing art when properly prompted. The example articles posted by Lambchop are quite fascinating, for an AI. We're in for a hell of a ride.

I'm not amazed by "AI" "Art".

Whether you invented this trope or not of "It's really all in how you prompt it" is growing tiresome.

"It's really all in how you prompt it" puts the responsibility back on the creative individual.


Do we want to play procedurally generated maps?

No.

 

[rusty_shackleford]

Do we want to play procedurally generated maps?

Do you prefer ubisoft maps?

The argument that procedurally generated is inherently worse isn't a very good one.

 

[Non-Edgy Gamer]

"It's really all in how you prompt it" puts the responsibility back on the creative individual.

True, though I didn't mean it like that.

You are the person using the tool. Modifying your use of the tool to improve the result should be common sense.

Do we want to play procedurally generated maps?

No.

As a big Daggerfall and roguelike fan in general, I'll have to disagree here. It just depends on the quality of those maps.

 

[grimace]

It just depends on the quality of those maps.

It just depends on the quality of the way you prompt it!!!

Are we stuck in a loop here?

What does it depend on? How is quality and craftsmanship judged to pass inspection?

Man needs to check the work of the AI algorithm procedurally generated map/art/story.

This AI stuff doesn't seem to make the job of man obsolete or less burdensome.

"AI becomes gamemaster in text based RPG. And handles everything."​

Fact Check

This title is provably false.

 

[rusty_shackleford]

It just depends on the quality of those maps.

It just depends on the quality of the way you prompt it!!!

Are we stuck in a loop here?

What does it depend on? How is quality and craftsmanship judged to pass inspection?

Man needs to check the work of the AI algorithm procedurally generated map/art/story.

This AI stuff doesn't seem to make the job of man obsolete or less burdensome.

"AI becomes gamemaster in text based RPG. And handles everything."​

Fact Check

This title is provably false.

A human DM also responds to the actions of the players, I fail to see how this is different.

 

[Non-Edgy Gamer]

It just depends on the quality of the way you prompt it!!!

No.

One is referring to whether or not I would like to play a game whose maps were procedurally generated by a developer or a program.

The other is referring to text generation where I am the one generating.

What does it depend on? How is quality and craftsmanship judged to pass inspection?

Again, you're talking about two different things. GPT and procedurally generated maps. One is relatively new and the other is decades old and in tons of games already.

If I am making something with a program, then it's my responsibility - whether I am a developer or someone generating text. If I am just a player, they I am judging the quality of the output and whether or not I like it only, since I have no control over the generation.

 

[RaggleFraggle]

“Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.”

 

[Sarathiour]

Do we want to play procedurally generated maps?

Do you prefer ubisoft maps?

The argument that procedurally generated is inherently worse isn't a very good one.

I'm not going to play either of them.

 

[rusty_shackleford]

Do we want to play procedurally generated maps?

Do you prefer ubisoft maps?

The argument that procedurally generated is inherently worse isn't a very good one.

I'm not going to play either of them.

But you play Rimworld...?

 

[Non-Edgy Gamer]

There is an element of hypocrisy and virtue signaling to every AI critic, it seems.

You can be against AI without being against all algorithms, or all automation, guys.

It's really hard to take these new Ted Kaczynski wannabes seriously when they're typing away on computers and using an internet that's increasingly powered by machine learning.

 

[Dadd]

AI produced crap is garbage and will always be garbage. The only problem is some of you don't seem to mind the taste of shit. You should rectify that.

 

[The_Sloth_Sleeps]

It just depends on the quality of those maps.

It just depends on the quality of the way you prompt it!!!

Are we stuck in a loop here?

What does it depend on? How is quality and craftsmanship judged to pass inspection?

Man needs to check the work of the AI algorithm procedurally generated map/art/story.

This AI stuff doesn't seem to make the job of man obsolete or less burdensome.

With the "AI" the choice of use boils down to do unsupervised/supervised.

First, generate utter turd content (text, art, etc) and put that straight into a game. The end result will be turd, maybe that's what's exciting some people here, the ability to churn out content quicker in exchange for quality and cost.

The other choice with using "AI" is to fix up generated content by hand, which I don't see a use case that it really takes any less skilled a person to do this.

The only counter argument I get to using "AI" is the broken record argument where pictures of garbage portraits/text of "AI" output is copy pasted into here, or a link with what might be possible in 2335AD.

 

[rusty_shackleford]

or a link with what might be possible in 2335AD.

Cutting edge AI used exclusively by multibillion dollar corporations in 2016:

Off the shelf consumer AI in 2022:

Careful not to overdose on that copium.

 

[Zeriel]

Can you point out the part that is amazing? Because it looks like the A""""I"""" is merely pretending to understand what you're saying without any background environment to relate what you're saying and respond appropriately.

Almost everything you see the last 10 years that people call "AI" is pattern recognition on steroids - it's very impressive for what it is, but don't think for a second that it 'learns' or 'understands', these are incredibly misleading terms that are for marketing purposes.

Or if you want 'intelligence', AI research has never, in its entire history, defined 'intelligence' in a rigorous way. Which is fine for most hucksters & companies, since you can basically call anything AI.

Anyway, this stuff could be used to spawn disposable, inconsequential NPC dialogue everywhere - the equivalent of Oblivion street talk, scripted so that the Mongolians shit-talk about the Hungarians and the Hungarians talk about potato pie, etc. It would make it way easier to churn out massive quantities of forgettable filler ... meaning exceptionally useful for, say, populating a gigantic metaverse that nobody wants to use. It's likely to make games bigger and more boring if anything.

Thank you. There are definitely applications for this tech, but it's so overblown in how people talk about it.

 

[Sarathiour]

Do we want to play procedurally generated maps?

Do you prefer ubisoft maps?

The argument that procedurally generated is inherently worse isn't a very good one.

I'm not going to play either of them.

But you play Rimworld...?

Yeah, and a few roguelike too. For an rpg, fps or platformer, good level design is obviously way more important than for some kind of sandbox, or a schmups for example. I thinks that should be pretty obvious why.

Do you really think procedurally generated Ubisoft style-map for whatever open world they are releasing next is going to be an improvement ?

 

[rusty_shackleford]

Do we want to play procedurally generated maps?

Do you prefer ubisoft maps?

The argument that procedurally generated is inherently worse isn't a very good one.

I'm not going to play either of them.

But you play Rimworld...?

Yeah, and a few roguelike too. For an rpg, fps or platformer, good level design is obviously way more important than for some kind of sandbox, or a schmups for example. I thinks that should be pretty obvious why.

Do you really think procedurally generated Ubisoft style-map for whatever open world they are releasing next is going to be an improvement ?

Ubisoft maps are handmade.

 

[Sarathiour]

That was precisely my point, they look incredibly bland despite being handmade. I have no doubt that it will be quite easy to get something that look barely distinct from ubi map using machine learning.

And it would just as much shit.