When the old school RuneScape wiki moved to a self hosted solution it was night and day the quality difference. I’d argue that OSRS (and probably regular RS) have some of the best wikis in gaming.
The technology was created to replace voice actors. That’s the actual purpose. Its very existence hurts their profession and benefits studios. You can not be a studio, use this technology, and claim to care about ethics, anymore than Amazon can claim to care about the workers as it invests in the machines to replace them.
No one is holding a gun to their head forcing them to us AI. They made a choice. There is no “ethical” way to cripple the livelihood of working class people for the benefit of your business. Just stop using the word.
It doesn’t matter if you compensate or get their approval, because the fact is the existence of the technology in the industry effectively compels all voice actors to agree to let it use their voice, or they can’t get work. It becomes a false choice.
If there was no financial benefit, if it truly made no difference in how much a studio pays in labor or the amount the artists make, there would be no reason for studios to want to use it.
Voiced characters that use generative AI in real time instead of prerecorded lines and a dialogue tree come to mind as an obvious use. How cool would that be, to be playing an RPG and ask any character any question you want and get an actual verbal answer? No way you can do that with voice actors.
Ever seen the game Vaudeville? It’s a fairly basic detective game but all the characters have their own LLM and AI voices. I bought it for the reason you described. I just had to see the technology in action and I can definitely see a future with generative text/voices in games.
It’s not perfect by any means but I think it’s a very cool approach to a detective game. There have been updates to it since I played that address most of the problems I had with it like characters forgetting past conversations and giving conflicting info.
I had spitballed an idea similar to this a few months back. Build the characters, world, and situations, and give the AI that information. Pick a few specific pieces of info the AI would have to tell you at specific times, basically to act as guide rails. Then, let the AI and the player just… Interact.
That’s pretty much Vaudeville. The only things you can do is click on locations and talk to people, each of whom has some bit of information you need to figure out.
It’s basically an experiment to see what works and what doesn’t with the idea. I appreciate that they kept the scope small (no quests, no WASD movement) and have been implementing changes as they discover the shortfalls (like the ones I’ve mentioned). If it ever does get released as a finished game, it’ll be more like a proof of concept for other games to build off of.
I find it to be very off putting that Baldur’s Gate 3 doesn’t have voice actors for the main character.
There are so many different races that would have different voices and different accents that it wouldn’t be financially viable to do that with voice actors either.
The only real ethical concern is around the training data. If all voices are compensated / actively consent to be used in an AI program, then this is just a tool. People losing jobs doesn’t really matter to an individual company. Industries change and technology advances.
So the real problem is they are using these types of tools, built of the skill of other voice actors, without properly compensating them or getting their consent.
What’s the point of bringing up “ethics?” The job only existed in the first place because of technology, and now people want to argue that there is a right or wrong aspect to it?
How about the poor candle makers or buggy whip manufacturers? Should we keep downgrading society just to keep a few “artists” happy?
Luddites were not anti-technology. They saw the progress of technology IN a primitive capitalist system and understood that technology would never benefit them, and always be used to subjugate them more.
If technology only benefits 0.1% of the world, and leads to the world dying, does it benefit humanity at all?
That’s kind of the point though isn’t it? It’s not the car’s fault we can’t afford the gas. We need to stop arguing about the ethics of using AI and start arguing about the ethics of the people using it unethically.
There is a person in that studio that suggested using AI, there is a person who gave the go ahead to do it. Those people need to be the problem, not the toy they decided to play with.
That's a very naive perspective though. We're not blaming the guns for gun violence, it's the people, but restricting access to guns is still the proven way to reduce gun incidents. One day when everyone is enlightened enough to not need such restrictions then we can lift them but we're very far from that point, and the same goes for tools like "AI".
Very easy time if it's about commercial use (well, at least outside of china). Companies need to have licenses for the software they use, they have to obey copyright laws and trademarks, have contracts and permissions for anything they use in their day to day work. It's the same reason why no serious company wants to even touch any competitor's leaked source code when it appears online.
Just because AI tech bros live in a bubble of their own, thinking they can just take and repurpose anything they need, doesn't mean it should be like that - for the most case it isn't and in this case, the law just hasn't caught up with the tech yet.
It’d be dead easy, actually. Don’t even have to actually ban it: For image generating models, every artist whose work is included in the training data becomes entitled to 5 cents per image in the training data every time a model generates an image, so an artist with 20 works in the model is entitled to a dollar per generated image. Companies offering image generating neural networks would near instantly incur such huge liabilities that it simply wouldn’t be worth it anymore. Same thing could apply to text and voice generating models, just per word instead of per image.
That said, this choice wasn’t actually a problem right?
I mean this game doesn’t use voice actors normally. If they used ai voice actors for this update only to represent the ai characters… isn’t that just appropriate?
Previously all characters in this game were represented only by text, so literally nobody is being replaced here.
Another way to think about it would be via representation. We get worked up when an ethnic character on screen is played by a different ethnicity, an actor in blackface for example. And in that vein using ai for organic characters could be seen as offensive, but using ai for ai characters would not. In contrast could we see using human voices for ai characters to be insensitive? That may sound far fetched, but this is sci-fi, the ai characters in the game are fully sentient and in their fictional universe would have rights, the whole point is to make the player think about what that means.
Well I guess I have my takeaway, I may consider boycotting any game that uses human actors for ai characters. Just get an ai actor… seriously.
Honestly, I’d argue that that’s exactly what AI should be for. Either being used by that one guy to give voices to his passion project because he can’t afford to hire voice actors, or to add a touch of the uncanny to an AI character.
Yes, as long as people keep focusing on fighting the technology instead of fighting capitalism, this is true.
So we can fight the technology and definitely lose, only to see our efforts subverted to further entrench capitalism and subjugate us harder (hint: regulation on this kind of thing disproportionately affects individuals while corporations carve out exceptions for themselves because ‘it helps the economy’)…
Or we can embrace the technology and try to use it to fight capitalism, at which point there’s at least a chance we might win, since the technology really does have the potential to overcome capitalism if and only if we can spread it far enough and fast enough that it can’t be controlled or contained to serve only the rich and powerful.
And I don’t make my own paints either when doing art. I still agree with the basic original point:
It is disappointing that we’re currently automating creativity far faster than manual labour. I’m angry that my art is getting automated away faster than my folding of laundry.
yea, see i just don’t like how we first automated creativity instead of like, idk, manual labor???
emphasis mine, but this is just incorrect. Technology has been reducing the need for manual labour (or rather increasing the amount of useful work done with manual labour) since the wheel and the plow.
It’s not; you’re just looking at the beginning of automating creativity when labor automation has been going on for over a hundred years. The introduction of new tech is always more disruptive than refining established tech. Besides which, VA is particularly sensitive to disruption because every VA does essentially the same job- one AI can be programmed to speak in thousands (millions?) of different voices, whereas one manual labor job doesn’t necessarily require the same actions as another.
Also it’s funny you complain about laundry, given how much doing laundry has been automated.
I have an idea for the practice that could help us better explore practical uses. Basically, a company may train an AI off an actor’s voice, but that actor retains full non-transferable ownership/control of any voices generated from that AI.
So, if a game is premiering a new game mode that needs 15 new lines from a character, but their actor is busy drinking Captain Morgan in their pool, the company can generate those 15 lines from AI, but MUST have a communication with the actor where they approve the lines, and agree on a price for them.
It would allow for dynamic voice moments in a small capacity, and keep actors in business. It would still need some degree of regulation to ensure no one pushes gross incentives.
I love the game and at times I couldn’t stop playing. The more it got patched the more issues I had with it and I haven’t played in a while. Still waiting for Falling Frontier to be released :)
I’ve played 1,300 hours of Destiny 2. I basically wrote my graduate thesis on how it uses psychological tactics to exploit its playerbase and how contemporary business models in gaming are harmful to consumers. Anyways, I gotta grind out my red borders before TFS drops.
I always think its the other way around. Some author writes a scary possibility about some topic that scared them but they don’t know a lot about. So like a book about a massive bedron impactor creates mini black holes that eats everything it touches. Book becomes popular and in ten years the LHC has some breakthrough but the zeitgeist was already established and people find all the reasons the cool ass tech is really going to be he worst thing ever.
It was crazy how swiftly media moved to present tons of reasons to hate AI.
It really made me realize how the people with this strongest opinions have been given those opinions by media that they don’t even realize is a form of media.
I want an actual real time strategy game. All popular RTSs are actually just about tactics and micro. I mean every SC2 guide will tell you that up to a very high level of play, if you’re just doing more you’ll be more efficient and win regardless of strategy. Why can’t you just set a standing order of “make unit x” or “make unit x while we have gas until we get to 50 of them”? That’s strategy. Having to tab back to a building and manually queue a couple of units every several seconds is just creating busywork for players, but thats what’s necessary and optimal for playing SC2 and most RTS games well
I love this concept; I had a friend from school viscerally defend SC: BW as superior to SC2 because in his words SC2 removed skill because of not having the unit select cap that BW did. That’s just less, as you put it, busywork, and then the player is more free to consider army compositions and positioning rather than drawing tons of rectangles. Removing more busywork in favor of actual strategy would be amazing.
Good point. I suppose I was combining the intended definition of micro as in issuing individual or otherwise sufficiently granular actions with the extra categorization of busywork, and indeed in that regard chess is pure micro.
There are types of time management which I think can still be interesting. For example, are you able to afford -- in the resources of time and attention -- optimally micro'ing this important fight? Or are you going to have to yolo it a bit so that you can do multi-task economic tasks at the same time?
Some (much?) of the problem is that (for better or worse) skilled players can and will squeeze the game to optimality in terms of win rate, and that tends to collapse viable tactical and strategic choices. Once those choices have been optimised (the game is largely "solved"), the main way to get better is by being faster, not by being smarter.
Hell, I should be able to upload an economic playbook with hundreds of rules like the one you described, and load it on game start. Then all I have to do is the actual unit movements.
Yep, take some ideas from single player colony management games.
It’s astounding how much you can “automate” when fully using the filters and rules options in vanilla Rimworld. Mods increase that exponentially. Granted, different genre, singleplayer, and pausable while you configure things.
I think the challenge is balancing that with the real time events you have to react to, so it doesn’t further compress the meta to an even smaller set of “optimal” options.
Supreme commander was what you describe. You setup your factory to make a unit or a set of units and repeatedly build them until canceled or not enough resources. You could zoom out to view the whole map. it was very much a strategy game and not really tactics or micro.
Rise of Nations (originally released back in 2003) had/has some interesting ideas to reduce some of the busywork:
Worker units will automatically try to gather/build nearby after a short (configurable) delay if they're not doing anything.
Cities (the main worker-producing structure) has a rally point option that's essentially "all nearby empty resource gathering", so you can queue a dozen workers and they'll distribute themselves as they're created.
Production buildings can be set to loop over their current queue, letting you build continually without intervention as long as you maintain enough resources each time the queue "restocks".
Units that engage in combat without being given an explicit target will try (with modest success) to aim for nearby units which they counter.
For the most part, none of the implemented options are strictly better than micromanaging them yourself:
You will always spend less time idling workers if you micromanage them yourself.
The auto-rally-point doesn't always prioritize the resources that you would if you did it yourself.
Queueing additional units is slightly less resource-efficient than only building one thing at a time.
Total DPS is higher if you manually micro effectively.
But the options are there when you need them, which I think is a a nice design. It doesn't completely remove best-in-class players being rewarded for their speed as a player, but does raise the "speed floor", allowing slower players to get more bang for their buck APM-wise, and compete a bit more on the strategy/tactics side of the game instead.
Because too much of SC2's design catered to the progamer crowd that liked that kind of stuff. They made some things easier from an APM standpoint but intentionally added more things to make the have not APM intense.
They really bet wrong on how popular that approach would be.
Why can’t you just set a standing order of “make unit x” or “make unit x while we have gas until we get to 50 of them”? That’s strategy. Having to tab back to a building and manually queue a couple of units every several seconds is just creating busywork for players
I agree completely. Related: have you considered turn based strategy games?
I feel like people dont understand, that the RT part in rts will always be the important part.
If you free up macro work, people will micro harder. WC3 got rid of most of the macro demand of SC and in consequence you will lose if you dont micro your units ik battle.
SC1 had build pipe lines and it wad still better to issue commands seperatley, because the player is more flexible.
A strategy is worthless if it csn be executed and the limits of execution create strategy.
Extraordinary pathing and all-select created the a-click deathball, that is one of the most boring ways to see, play and lose to.
That is not true, at least in Age of Empires 2 which is the RTS I’m most familiar. Have a look at the limited viper series to see a good player destroy using only 60 APM. If you make good decisions, you don’t need to click as much.
Why can’t you just set a standing order of “make unit x” or “make unit x while we have gas until we get to 50 of them”?
Because while this will make casuals that will play the game for 3 hours and drop it happy, the typical RTS fans will not enjoy this. There is a trade off between queuing a lot of units and having more resources available for other techs. Having units auto produce without any disadvantage is just kind of boring. Then you are just watching the game, not really playing it.
Maybe you should try turn based strategy, if you don’t like real time strategy. In the later, like the name implies, time is the most important resource. You don’t need a lot of clicks, but you need to use it wisely.
I get that record sessions are a huge hassle and simply paying VAs per AI-generated voice line is easier for everyone, but it somehow makes Paradox look a little careless to me.
Stories like these also set a precident. This is what voice ‘acting’ will be like for a moment before it becomes effectively eliminated because voice libraries will become diverse enough quickly and there will be no need for a single more voice actor to be included. It seems like VAs are basically forced to sell their voice to AI companies quickly to at least make a quick buck before they never get a job again.
There’s probably no stopping it, but that made this read all the more frustrating to me.
This is what (modern) voice acting has always been.
Actually read a few interviews with professional VAs or watch their streams if they do that. Two VAs actually interacting with each other and reacting is almost unheard of outside of very specific productions (and mostly are done as a stunt for some BTS footage). They read a dozen different takes of every line and go through like five different scripts worth of dialogue. And then they do “efforts” that are just general grunts and emoting that are used for the moment to moment gameplay and to pad out a scene that had heavy rewrites. It is why so many professional VAs can stream “their” games… because they genuinely have no idea what is going to happen.
Paying to train a limited use model off of a specific VA (or even a group of VAs) is the “logical” extension of that. And, arguably, it is a “good” one (with some MASSIVE caveats). Everyone lost their god damned mind over that FPS that came out last year where the announcer was (allegedly?) a model trained off of a VA. But it also meant that you could have stuff you would never have had otherwise. Nolan North isn’t going to get a paycheck to sit in a booth all day commenting on random matches. But a model that can read out a team’s name and string together different reactions? That is actually really cool and WAY better than the traditional sports game approach of “The Champion! just went through… A Table!”*
Like almost everything AI? The key is to focus on creators’ rights and control what can and can’t be used as training data. Because the genie is out of the bottle and ain’t going back in. But if we can protect the rights of what goes into training data? Then people are still paid for their effort/creation.
Do I think this was done “ethically”? I don’t know. But with everything Paradox has done in the past few years? I assume “not in the slightest”. But the concept is sound and one that we need to standardize sooner than later.
Of course, we also need UBI so that people’s lives aren’t tied to their jobs but that is a bigger mess.
*: Also, if you don’t think those aren’t already stitched and blended together with most of the same tech then I have a bridge to sell you
I’ll also add on that there are very good reasons to pay for models based on VAs. Brendan Fraser infamously permanently-ish hurt his vocal cords because of the performance that were expected of him in his prime. Same with a lot of VAs (I think David Hayter is one?) who basically need to smoke a pack a day when they are “in character” to get the right gravely voice. And while Stephanie Beatriz played it smart and made sure her “Rosa” voice was something she could maintain, a lot of actors and actresses basically can’t be the character they are famous for because it is killing them.
And pulling a solution out of my ass that is surely missing important aspects of the industry?
if I just want Nolan North or Felicia Day to voice a character then I buy the use of their model from their agency and am charged based on how much dialogue they have in a given game. If I want to use them as a character going forward (so what ANet tried with Felicia before they realized she was too expensive and decided to give Zojja permanent brain damage so she wouldn’t ever have dialogue again)? I can pay by line at a much cheaper rate.
But if I want Nolan North to do a voice that isn’t just Drake? Then I am paying him to train a new model and it gets a lot more expensive. And I can pay more to “own” that training data with the same caveats regarding future use. The main idea being that I want to make sure my Nolan North performance doesn’t end up in a competitor’s game next week.
Have any of the involved voice actors confirmed the claims made in the article? I’ve seen multiple articles on this game, and the only quotes are from the game’s director.
So far I’ve only seen one side claim this is ethical. That’s not enough.
I’m not really up-to-date on voice synthesis. Have we reached the point where we can get enough training data from just a handful of voice actors to train a model of this quality?
Or is this a case of them using those voice actors for fine-tuning a pretrained model and just being quiet about that?
Yeah, if Mozilla’s goal is 1200 clips/day and 2400 validations/day then I have a strong suspicion that Stellaris uses a pretrained model and there are no royalties for the people whose voices were used for the pretraining. Not that it would be feasible to spread royalties among that many people in the first place.
What could point against that suspicion though is that Stellaris doesn’t need a “perfect” model so maybe they can get away with much, much less. After all the whole gimmick is that it is in-universe AI. A (near-)flawless model would be (near-)indistinguishable from a regular voice actor. Then there would’ve been no need to hire a bunch of voice actors to train an AI in the first place.
Assuming that it is pretrained -> finetued though, the only hope is that those initial files were donated willingly and not scraped somewhere. Otherwise their “ethical” argument goes out the window.
You’re allowed to like and enjoy things, even bad things… Hell, theres entire fandoms around liking bad things (like B-Movies), that doesnt make them less fundamentally bad. and it doesnt make you wrong for liking them.
I’m wondering if better AI could save this genre. I always hated the fragility of any soldiers I wasn’t actively controlling, having idle workers, workers trying to chop wood in the middle of enemies, etc.
If the computer can take your high level commands but also put out logical low level ones, and maybe also punish high APM, it might restore some of the moderate-paced feel of the game.
It’s a question of whether to reward a player that can see that the opponent is using rock, take a step back, start building paper, and send them out even if they take time doing it; versus a player that just super-optimizes building an army of rock to send against armies of paper, and give them the best chance of winning by perfectly kiting every attack on the field.
There’s certainly an argument that some groups would like the tournament of APM, but I think a lot of people didn’t bother with high level StarCraft because they saw Koreans clicking 15 times a second and figured they can’t keep up. It’s like how fighting games work to demonstrate they’re not rewarding button mashing.
pcgamer.com
Aktywne