Interesting read on AI powered MMO platform under development

mikeb0817

Unleashed is the name - AI and proc gen enable creation of big games much faster, is the gist of it.

“"Kingdoms of Amalor never made a dime, but it cost $133 million in world building. And that’s where the cost is: content generation. We can do it in 10% to 20% of the time, and the same for cost,” Pereira said.""

https://venturebeat.com/games/unleashed-is-making-games-that-bring-people-together-the-deanbeat/

mikeb0817

I also don't ever remember an R.A. Salvatore MMORPG being in development. Would have been nice.

Angrakhan

The problem with using generative AI to create games is the same problem as all generative AI. That is it works by copying other people's work and creating various "mashups" of other people's work and calling it new. Let's forego the ethics discussion of putting humans out of work using this method which also is clearly happening here and focus instead on what this mashup technique means. It means no matter how many times you run your generation algorithm what it's going to spit out is derivative. It will never accidentally innovate. You might say that's all humans do, but humans do indeed innovate. Someone made EQ for WoW to copy. Someone made Doom for a billion shooters to copy. Someone made the original Dune RTS for Warcraft to copy. Without that innovative potential you're just signing yourself up to be an also-ran. 

How many of us can look at art and instantly say "that's AI"? It's pretty obvious right? It all got the same generic derivative look. 

Now imagine a cavalcade of MMOs generated with AI. You think what is out now is derivative tripe? This will be copy and paste on a whole new level. 

BUT THEY CAN DO IT IN 10% OF THE TIME!

Ok but is that helpful for anything other than the developer's bottom line?

Sovrath

mikeb0817 said:

I also don't ever remember an R.A. Salvatore MMORPG being in development. Would have been nice.

It was huge news as was the involvement of Curt Schilling as it was his game company. He ended up making a deal with Rhode Island in order to get a loan but develop the game there. All his company made was Kingdom of Amalur as they ended up bankrupt.

Some say they made Amalur to get money to help fuel the company and some say they always intended two games.

mikeb0817

Angrakhan said:

The problem with using generative AI to create games is the same problem as all generative AI. That is it works by copying other people's work and creating various "mashups" of other people's work and calling it new. Let's forego the ethics discussion of putting humans out of work using this method which also is clearly happening here and focus instead on what this mashup technique means. It means no matter how many times you run your generation algorithm what it's going to spit out is derivative. It will never accidentally innovate. You might say that's all humans do, but humans do indeed innovate. Someone made EQ for WoW to copy. Someone made Doom for a billion shooters to copy. Someone made the original Dune RTS for Warcraft to copy. Without that innovative potential you're just signing yourself up to be an also-ran. 

How many of us can look at art and instantly say "that's AI"? It's pretty obvious right? It all got the same generic derivative look. 

Now imagine a cavalcade of MMOs generated with AI. You think what is out now is derivative tripe? This will be copy and paste on a whole new level. 

BUT THEY CAN DO IT IN 10% OF THE TIME!

Ok but is that helpful for anything other than the developer's bottom line?

I don't expect them to reinvent the wheel - but look at what proc gen has allowed No Man's Sky to do since the launch disaster. It's one of my favorite games of all time now. If they had built WoW with the Unleashed platform, described the way it is, Blizzard could turn out a copy in the style of StarCraft quick and cheap - with all the same controls and UI elements we're familiar with - in record time. I would absolutely buy that game. 

Quizzical

AI generated content should be thought of as a particular type of procedurally generated content.  And not a particularly good type of procedurally generated content.

Quizzical

mikeb0817 said:

Angrakhan said:

The problem with using generative AI to create games is the same problem as all generative AI. That is it works by copying other people's work and creating various "mashups" of other people's work and calling it new. Let's forego the ethics discussion of putting humans out of work using this method which also is clearly happening here and focus instead on what this mashup technique means. It means no matter how many times you run your generation algorithm what it's going to spit out is derivative. It will never accidentally innovate. You might say that's all humans do, but humans do indeed innovate. Someone made EQ for WoW to copy. Someone made Doom for a billion shooters to copy. Someone made the original Dune RTS for Warcraft to copy. Without that innovative potential you're just signing yourself up to be an also-ran. 

How many of us can look at art and instantly say "that's AI"? It's pretty obvious right? It all got the same generic derivative look. 

Now imagine a cavalcade of MMOs generated with AI. You think what is out now is derivative tripe? This will be copy and paste on a whole new level. 

BUT THEY CAN DO IT IN 10% OF THE TIME!

Ok but is that helpful for anything other than the developer's bottom line?

I don't expect them to reinvent the wheel - but look at what proc gen has allowed No Man's Sky to do since the launch disaster. It's one of my favorite games of all time now. If they had built WoW with the Unleashed platform, described the way it is, Blizzard could turn out a copy in the style of StarCraft quick and cheap - with all the same controls and UI elements we're familiar with - in record time. I would absolutely buy that game. 

AI could allow Blizzard to crank out 100 bad knock-offs of StarCraft very quickly.  All of them would be vastly inferior to the StarCraft that they actually made, as well as to most (likely all) of the other RTS games that reputable studios have made.  Why would you want to play a terrible game over a good one?

Making it really cheap to make terrible games is a bad thing, not a good thing.  If they don't get a handle on this, then game sites like Steam, GOG, and Itch could soon be flooded with so many awful, AI-generated games that it will be nearly impossible for anyone to find the actually good games made by new, indie developers.

mikeb0817

Quizzical said:

mikeb0817 said:

Angrakhan said:

The problem with using generative AI to create games is the same problem as all generative AI. That is it works by copying other people's work and creating various "mashups" of other people's work and calling it new. Let's forego the ethics discussion of putting humans out of work using this method which also is clearly happening here and focus instead on what this mashup technique means. It means no matter how many times you run your generation algorithm what it's going to spit out is derivative. It will never accidentally innovate. You might say that's all humans do, but humans do indeed innovate. Someone made EQ for WoW to copy. Someone made Doom for a billion shooters to copy. Someone made the original Dune RTS for Warcraft to copy. Without that innovative potential you're just signing yourself up to be an also-ran. 

How many of us can look at art and instantly say "that's AI"? It's pretty obvious right? It all got the same generic derivative look. 

Now imagine a cavalcade of MMOs generated with AI. You think what is out now is derivative tripe? This will be copy and paste on a whole new level. 

BUT THEY CAN DO IT IN 10% OF THE TIME!

Ok but is that helpful for anything other than the developer's bottom line?

I don't expect them to reinvent the wheel - but look at what proc gen has allowed No Man's Sky to do since the launch disaster. It's one of my favorite games of all time now. If they had built WoW with the Unleashed platform, described the way it is, Blizzard could turn out a copy in the style of StarCraft quick and cheap - with all the same controls and UI elements we're familiar with - in record time. I would absolutely buy that game. 

AI could allow Blizzard to crank out 100 bad knock-offs of StarCraft very quickly.  All of them would be vastly inferior to the StarCraft that they actually made, as well as to most (likely all) of the other RTS games that reputable studios have made.  Why would you want to play a terrible game over a good one?

Making it really cheap to make terrible games is a bad thing, not a good thing.  If they don't get a handle on this, then game sites like Steam, GOG, and Itch could soon be flooded with so many awful, AI-generated games that it will be nearly impossible for anyone to find the actually good games made by new, indie developers.

I think you're misunderstanding. 


If WoW had been originally made on the platform, using proc gen and AI for art assets would have allowed them to make World of StarCraft quickly, and it would be familiar to anyone who had played WoW since it would be a copy of the game will a new skin. Certainly if they could do that in 10% of the time, and turn out a product that is exactly like WoW but StarCraft flavor, you think that would be a bad thing?

I think some of you are bored and enjoy playing devil's advocate just to continue a conversation. I doubt you even read the article.

Dattelis

The problem isn't the potential of the tool (which that's all A.I. is, its a tool), but the utilization of it and essentially the consolation of fields in bigger companies at least. Having A.I. doesn't mean that there will automatically be more products, it means the company operate leaner and make more overhead first. So you essentially kill off positions/jobs when something like A.I. gets integrated further. Or do people seriously think CEOs and board members really care about product quality?

Quizzical

mikeb0817 said:

Quizzical said:

mikeb0817 said:

Angrakhan said:

The problem with using generative AI to create games is the same problem as all generative AI. That is it works by copying other people's work and creating various "mashups" of other people's work and calling it new. Let's forego the ethics discussion of putting humans out of work using this method which also is clearly happening here and focus instead on what this mashup technique means. It means no matter how many times you run your generation algorithm what it's going to spit out is derivative. It will never accidentally innovate. You might say that's all humans do, but humans do indeed innovate. Someone made EQ for WoW to copy. Someone made Doom for a billion shooters to copy. Someone made the original Dune RTS for Warcraft to copy. Without that innovative potential you're just signing yourself up to be an also-ran. 

How many of us can look at art and instantly say "that's AI"? It's pretty obvious right? It all got the same generic derivative look. 

Now imagine a cavalcade of MMOs generated with AI. You think what is out now is derivative tripe? This will be copy and paste on a whole new level. 

BUT THEY CAN DO IT IN 10% OF THE TIME!

Ok but is that helpful for anything other than the developer's bottom line?

I don't expect them to reinvent the wheel - but look at what proc gen has allowed No Man's Sky to do since the launch disaster. It's one of my favorite games of all time now. If they had built WoW with the Unleashed platform, described the way it is, Blizzard could turn out a copy in the style of StarCraft quick and cheap - with all the same controls and UI elements we're familiar with - in record time. I would absolutely buy that game. 

AI could allow Blizzard to crank out 100 bad knock-offs of StarCraft very quickly.  All of them would be vastly inferior to the StarCraft that they actually made, as well as to most (likely all) of the other RTS games that reputable studios have made.  Why would you want to play a terrible game over a good one?

Making it really cheap to make terrible games is a bad thing, not a good thing.  If they don't get a handle on this, then game sites like Steam, GOG, and Itch could soon be flooded with so many awful, AI-generated games that it will be nearly impossible for anyone to find the actually good games made by new, indie developers.

I think you're misunderstanding. 


If WoW had been originally made on the platform, using proc gen and AI for art assets would have allowed them to make World of StarCraft quickly, and it would be familiar to anyone who had played WoW since it would be a copy of the game will a new skin. Certainly if they could do that in 10% of the time, and turn out a product that is exactly like WoW but StarCraft flavor, you think that would be a bad thing?

I think some of you are bored and enjoy playing devil's advocate just to continue a conversation. I doubt you even read the article.

Do you mean exactly the same underlying mechanics as WoW, but only with art assets changed to look like StarCraft?  What about quest text and lore?

mikeb0817

This is Haven. It's the proof of concept for the platform.
 https://hardforum.com/threads/haven-new-mmo-by-unleashed-games.2033427/
Anything built on the platform that turns out to be successful can easily be modified. I imagine many different base groups of game elements that can be defined to generate endless creative variables for each individual aspects you want to tweek. by a bunch of different proc gen (I'm thinking landscapes) and AI (ART - things like character model / item skins / kind of anything along those lines).

They've said a few times these are new tools to help developers create more, and do it faster - not to replace artists.. I suppose that anything gamers seem to like that is built on the platform could be used as the base model. Then you could pick different aspects you want to spice up and give it to the platform use the original recipe and spin out a new game. That's where my comparison from WoW  (successful model) and a new IP (StarCraft) that could have been spun out of it. In fact they could have taken all the core aspects of WoW that the devs think gives WoW the special sauce, make a bunch of other shit variable,.and let it help you make a whole host of new game - keeping to your core vision of what people are drawn to in your game. 

I probably need to edit for clarity later but tired.

Scot

My main concern is not quality, that's only secondary. More than that quantity is a bigger issue, we already have seen the effect of the ultra proliferation of games on the gaming market, the baulk of multiplayer games don't get past the two month point before the population nose dives. If studios can double, treble, quadruple their output that nose dive might be no more than two weeks.

The only thing that would stop that would be the price of each game. If enough were F2P or enough turned up on game passes, games may end up scrambling to get two weeks of a gamers time. This feeds into the super casual ding, ding mentality, which is dangerous for the industry. If you keep giving the consumer more and more he will value what you are making less and less.

mikeb0817

Think about it differently then, Scot. If Haven turns out good (or from my example - something popular like WoW was built on the platform) the devs are set up to win when it comes to ease of producing regular expansions. Don't think of them turning out small, frequent, content drops just because the speed of development has increased 80-90%, but rather imagine if your expansion for your favorite game could now include 80% more of the content you want in EACH expansion. Bigger, filled with more of the content you already love from the game.

The tool has amazing potential in the right hands.

Brainy

AI is not bad, its just a tool.  Nothing says you cant have the AI do 80% then let humans finish it.  It all just depends on who completes the finished product and how thats done.

This is no different than any technology.  Tech can speed up a process, its not by itself bad or good.  Its all about who uses it and how they use it.

Will there be a bunch of cheap knockoffs.  Probably.  But then again, it could allow some people to make something better than they otherwise could have.

You can only go so far with stone tools.

mikeb0817

Right. I think these tools will produce both - shallow cheap products, and really cool products that a small studio or individual couldn't do before. We will have to sift through trash to find the gems, obviously. I'm just trying to look at the positive end of the spectrum, rather than the gloomy end.

cameltosis

There's a whole bunch of things in the OP's article that has me confused!

1) I own Kingdoms of Amalur: Re-reckoning.

The article kept saying that the studio just burned $133m on this project, but I own it, I've played it, I enjoyed it. Are there two games by the same name? Was Curt's studio developing a sequel that never released, and that's why they lost money? The article is very unclear on this.

2) R.A. Salvatore's MMO was/is Neverwinter

Granted, I don't believe it is actually massively multiplayer, they had to scale back the multiplayer aspect as they couldn't get it working right. But, its a multiplayer game based on Salvatore's work, and I think Salvatore even wrote a bunch of new books to accompany the game (Gauntlrygrim?). He was definitely closely involved during development. The game is still going.

3) Whilst content costs a lot of money to make, MMORPGs fail because of bad gameplay

I'm sure generative AI can reduce the cost of content production, even if what it produces is very derivative. But that won't help make your game successful. Gameplay remains king, if your gameplay isn't engaging, players won't stick around long enough to form those social bonds that increase retention. But AI can't do shit for improving gameplay, that still requires highly intelligent, knowledgable, skilled designers and developers.

4) Procedural Generation is better than AI for game development

Procedural generation is a proven technology that keeps getting better, is fully under the developers control, isn't about to get completely screwed by copyright laws, and can be scaled up nicely. SWG is an early example of procedural generation in MMOs, allowing that game to have massive worlds compared to it's competitors. Now Raph is taking the next step with the tech for Star's Reach.

Theocritus

Angrakhan said:

The problem with using generative AI to create games is the same problem as all generative AI. That is it works by copying other people's work and creating various "mashups" of other people's work and calling it new. Let's forego the ethics discussion of putting humans out of work using this method which also is clearly happening here and focus instead on what this mashup technique means. It means no matter how many times you run your generation algorithm what it's going to spit out is derivative. It will never accidentally innovate. You might say that's all humans do, but humans do indeed innovate. Someone made EQ for WoW to copy. Someone made Doom for a billion shooters to copy. Someone made the original Dune RTS for Warcraft to copy. Without that innovative potential you're just signing yourself up to be an also-ran. 

How many of us can look at art and instantly say "that's AI"? It's pretty obvious right? It all got the same generic derivative look. 

Now imagine a cavalcade of MMOs generated with AI. You think what is out now is derivative tripe? This will be copy and paste on a whole new level. 

BUT THEY CAN DO IT IN 10% OF THE TIME!

Ok but is that helpful for anything other than the developer's bottom line?

ITs kind of the same thing that happened in music....THey took the human element out and replaced it with technology....Sure it is moe accurate than a human, but doesnt have the same sound....In our drive for perfection, we left something behind: human innovation. I dont like the way it turned out in music, and I doubt I like the way it will turn out in games.

[Deleted User]

The user and all related content has been deleted.

Xiaoki

cameltosis said:

There's a whole bunch of things in the OP's article that has me confused!

1) I own Kingdoms of Amalur: Re-reckoning.

The article kept saying that the studio just burned $133m on this project, but I own it, I've played it, I enjoyed it. Are there two games by the same name? Was Curt's studio developing a sequel that never released, and that's why they lost money? The article is very unclear on this.

2) R.A. Salvatore's MMO was/is Neverwinter

Granted, I don't believe it is actually massively multiplayer, they had to scale back the multiplayer aspect as they couldn't get it working right. But, its a multiplayer game based on Salvatore's work, and I think Salvatore even wrote a bunch of new books to accompany the game (Gauntlrygrim?). He was definitely closely involved during development. The game is still going.

3) Whilst content costs a lot of money to make, MMORPGs fail because of bad gameplay

I'm sure generative AI can reduce the cost of content production, even if what it produces is very derivative. But that won't help make your game successful. Gameplay remains king, if your gameplay isn't engaging, players won't stick around long enough to form those social bonds that increase retention. But AI can't do shit for improving gameplay, that still requires highly intelligent, knowledgable, skilled designers and developers.

4) Procedural Generation is better than AI for game development

Procedural generation is a proven technology that keeps getting better, is fully under the developers control, isn't about to get completely screwed by copyright laws, and can be scaled up nicely. SWG is an early example of procedural generation in MMOs, allowing that game to have massive worlds compared to it's competitors. Now Raph is taking the next step with the tech for Star's Reach.

1 - there was a Kingdom of Amalur 2 planned but it never started production. The game that would become KoA was already in development by Big Huge Games, 38 Studios saw what they were working on and liked it so they bought Big Huge Games for around $70 million. They had to change what was already made into KoA then finish the game.

They bought a company and a game already in production to serve as a quick introduction to the world and get people excited for the upcoming MMO. But, if they had spent all of the money on the MMO then they definitely would have had enough to finish the MMO.

2 - RA Salvatore was hired to write the lore, characters and story of KoA and Project Copernicus, which was the MMO. The character designs were done by Joe Madureira (comic book artist). 

RA Salvatore didnt have much to do with Neverwinter at the start, mainly just coordination between him and Cryptic about which characters are going to be killed, when and how. He later wrote a Drizzt quest line for the game.

3 - Gameplay is subjective. Some people would say action combat is bad gameplay and tab target is good gameplay and some people would say the opposite.

4 - generative AI has been used in video game development for years now. But, most people didnt know "Generative AI" existed until ChatGPT made it a buzzword.

mikeb0817

""""4) Procedural Generation is better than AI for game development

Procedural generation is a proven technology that keeps getting better, is fully under the developers control, isn't about to get completely screwed by copyright laws, and can be scaled up nicely. SWG is an early example of procedural generation in MMOs, allowing that game to have massive worlds compared to it's competitors. Now Raph is taking the next step with the tech for Star's Reach''.""""

When applied to game dev, I look at the way A.I. is applied to being very similar to proc gen. Proc gen typically is used for landscape and world generation, where I see A.I. being most useful is with character models and textures. Sets of armor, enemy mobs, any item with a visual representation. Graphic enhancement being the main best use imo. Both will advance quickly I think, to the point where in a few years what we see today would be like comparing an iPhone to the old Nokia I played snake on. Not the best analogy, but you get it.

EDIT: sorry, not a big fan of "quoting" feature.

BrotherMaynard

Say what you will about AI-generated music, no AI will ever come close to the human genius, such as the timeless classics like Mulligan and O'Hare.



If I ever see AI produce something even remotely approaching the quality of The Unbearable Pyjama Party, I will drop my sandwich.

GrimDogGaming

IF you can provide me with a complete AI dev team in a box, I will provide an mmorpg you'll actually play and enjoy.

Robokapp

Angrakhan said:

 It means no matter how many times you run your generation algorithm what it's going to spit out is derivative.

over a large-enough sample size, that's not an issue.

if an AI took every written text ever written and assembled it into new books, taking one sentence at a time, nobody would go "oh, this is familiar".

it'd look new to every consumer.

[Deleted User]

The user and all related content has been deleted.

Blueliner

I personally find it hilarious that the tech companies are actively developing the very thing that will make them obsolete

[Deleted User]

The user and all related content has been deleted.