I've developed a patent-pending technology for automatically verifying whether content is human-generated or AI-generated, which I call Miles. You can read about it at https://miles.dails.net
I'm interested in working with game developers and itch.io platform engineers to build capabilities and integrations with it to support human gamers and game designers. If anyone has any interest or ideas or relevant connections, please let me know!
The thing you describe does not verify that something is human made or ai made.
The only thing this does, is verify membership of a linked network of posters. Like a friend of friend in social media. And you verify if that friend's friend is also member of a separate exclusive meta social network on top of having an account on that social media platform you are trying to read the post on. Effectively whitelisting members.
And that exclusive meta social network has the membership entry bar of: please promise to not post ai content.
This will not catch retweets of questionable posts or any other kind of social media sharing.
Frankly, this is never ever going to work, not for the intended purposes. Just to put out a ball park number for the scale you need to consider:
Facebook removes over 10 000 000 fake accounts per day. So if you want this to fly on social media you have to prepare for that. And you have put examples on your site, how this is supposed to work on social media by just not showing the fake ai posts. Bad actors would find ways to exploit this network and it gets worse. The added trust of your friend's friend trust zone, would make you believe the fake posts even more, in the time frame it takes to remove the bad actor. And if you regularly have to remove bad actors, the whole concept of trusting the posts of members of the network is made obsolete.
For game publishers, if you use the concept of trust, there is just that. You do not need an added layer of tech for that. Especially not a place like Steam. If they ask you to disclose your ai usage and they catch you deliberatly lying and doing false advertisement, they will give you trouble. And big game studios will just not buy assets from an unknown asset creator that published a hundred of those on Itch. For that matter, not even small studios will do that. It is hobby developers that do.
And in my opinion, no one cares if a hobby dev uses ai.
The current trust system on Itch is, that you need to disclose the assets you publish. So if you look in the info box on an asset page, it will say ai generated or it will say no ai used.
Now, what about people lying? Same problem as in your proposed solution. You need to remove those. If you find proof that someone is lying, report it to Itch and let them deal with the fraudster.
I just do not see the problem (for games context). This is not something where people get rich quickly or can scam millions, or try to influence politics. There is too little incentive to lie.
Do not get me wrong, a method of blocking bots and fake posts on social media would be wonderful. But your method is just shifting the sign-up process from a platform to yet another platform. It is not dealing with the problem of bad actors or solving the problem of securing the sign-up process for social meda and it does not deal with people posting ai slop or fake news.
And using such a method for ai games, is just overengineering. If you need trust in the actors, just trust them to not tell lies about the ai status of their creations.
On your site you used a game published by Paradox Interactive. This is how a Steam ai disclosure reads for one of the games they published:
AI Generated Content Disclosure
The developers describe how their game uses AI Generated Content like this:
We leveraged AI-generated content to enrich our game with elements that would have been otherwise challenging to include. Specifically, we used AI for the AI character voiceover in different languages, a feature made possible thanks to this technology. In this case it made even more sense since the character is actually an AI. Otherwise, all the core art and visual elements of the game have been designed by human talents.
(Steam does not care about ai assisted programming. Only about things that influence the content directly. Not methods how to code the same function faster.)
Thank you for the feedback! This is very thoughtful! I have some additional perspective for you.
"The thing you describe does not verify that something is human made or ai made."
That's correct! The goal of the tool is not to replace human judgment with machine judgment, it's to scale human judgment.
"This will not catch retweets of questionable posts or any other kind of social media sharing."
Yes it will! The signer's metadata travels with the content unless deliberately removed, so if you, a person I trust, share something signed by a person I don't, it won't register as verified and it can be marked or filtered out at the application layer.
"So if you want this to fly on social media you have to prepare for that"
No I don't! Social media platforms all already have account integrity programs, there's no need to invent my own. Also, see immediately below.
"And if you regularly have to remove bad actors, the whole concept of trusting the posts of members of the network is made obsolete."
But you don't have to! A bad actor making an account must convince others to mark him as trusted and stay that way. If someone is going around marking shady accounts as trusted, other users will remove that enabling user from their trust network by removing trust links or adding distrust links. Assuming selfish behavior for every user, the trust network heals itself. You can make 10,000 fake accounts, but without getting others to mark you as trusted, it gets you nowhere
"For game publishers, if you use the concept of trust, there is just that. You do not need an added layer of tech for that"
Sure you do. Trust, but verify. Would you pay a credit card bill that wasn't itemized because you otherwise trust the bank? Would you pay for a meal without looking at the receipt? We add a tech layer to trust all the time, you're just so used to them that they're invisible to us.
"If you find proof that someone is lying, report it to Itch and let them deal with the fraudster"
...who will promptly change their name or account and go on defrauding people. If trust follows you, reputation matters.
"If you need trust in the actors, just trust them to not tell lies about the ai status of their creations."
'Trust be, bro' doesn't cut it anymore. We're past that.
"And using such a method for ai games, is just overengineering."
It is not. In fact, it's designed to be nearly transparent; if the software you're using knows how to read the file format, the end user and the creator have to set it up once (5 minutes) and never think about it again. That's the beauty of computers.
"The developers describe how their game uses AI Generated Content like this:"
Sure, but that's just Trust Me, Bro. How much of it is AI generated? Can you verify that? Which voice actors provided which lines? Did you notice the AI-generated content that sneaked through QA in Claire Obscure? You know what would have caught it?
tl;dr your approach cannot detect and cannot prevent. You merely shift the housekeeping from the social media platform to your network.
(social media sharing) so if you, a person I trust, share something signed by a person I don't, it won't register as verified
Most "posts" are not original content, but shared. If your solution blanket blocks almost everything, it will not be very useable in a social media situation.
A bad actor making an account must convince others to mark him as trusted and stay that way.
I hope you do realize how trivial it is to circumvent this. You are probably a nice and honest person. So you might lack the imagination a bad actor will have and the things a bad actor will come up with.
Would you pay a credit card bill that wasn't itemized because you otherwise trust the bank? Would you pay for a meal without looking at the receipt? We add a tech layer to trust all the time, you're just so used to them that they're invisible to us.
This is not the same type of trust. You are mixing trust in good intention with trust in acting without errors. And you are mixing up whom I would not trust. A bank usually is not out to scam customers (not in this way at least). But errors could happen, mostly because the billing is done by external third parties. So if I distrust a non itemized bill, it is not because I do not trust my bank, but because errors can happen and there are people scaming people with fake charges.
What are you refering to with "them" and "they" here? Are you talking about tech layers of trust or some concept like that? If so, that layer is not technical in nature, but the actual law. We generally trust that people obey the rules of the law and the rules of society. The problem is the exceptions, and you have to guard against those bad actors.
I get the concept of creating a gated community of nice people. But this does not scale up.
Creating a big tech layered network of people trusting people that trust people that can be trusted to be trusted by people trusting people that trust people will eventually approach a society model. And we already know that this does not work in actual society. You will have bad actors in that trust network that do damage before you can remove them. And if you cannot prevent the acts, the whole concept of the trust network is as good or as bad as the account integrity program of that network.
Social media platforms all already have account integrity programs, there's no need to invent my own
I think I just told you why you do need this, and that also implies why your's need to be better than the account integrity of the actual social media platform. Think about it. If your verification process is worse than that of, say, facebook, what is preventing the 10 million fake accounts per day to just infiltrate your network? We just established that you either do not have an integrity program at all or that your program is worse or equal than the actual account verification of a multi billion dollar company.
So to be of any use, you do need to offer something that the other platforms can not offer. You describe it as trust. But trust in what? Trust that members of the network do not post bad stuff. But for that, you need to have trust that these persons are persons and not fake accounts. How do you establish that? By removing the bad ones after the fact What you call self healing. But if you have to remove them, that means, they were able to make an account and enter the network.
The scammers and bad agents do not care if you remove them. You need to remove them before they can deliver their payload. If you remove them after, your network has achieved nothing, except make everything more complicated and mislead people into trusting the fake members of the network, undermining the trust in that network.
And this is the biggest issue with your concept: your approach cannot detect and cannot prevent. You merely shift the housekeeping from the social media platform to your network.
who will promptly change their name or account and go on defrauding people. If trust follows you, reputation matters.
'Trust be, bro' doesn't cut it anymore. We're past that.
I do not know what that bro phrase means. And yes, reputation matters. That is why you do not buy assets for a non hobby project from an unknown source.
(overengineering) It is not.
You propopse a method that would solve the issue of bot accounts, scammers, fake news and whatnot and want to use it to solve the purity issue of human made vs. ai made content in video games.
I do call that overengineering. And by citing the Claire Obscure thing, you have pointed out that your solution is not needed for that scenario. Reputation does matter. And trust me, because if I am cought lying I will face consequences on the platforms I publish is real and does cut it.
Sure, but that's just Trust Me, Bro. How much of it is AI generated? Can you verify that? Which voice actors provided which lines? Did you notice the AI-generated content that sneaked through QA in Claire Obscure? You know what would have caught it?
We are not talking about allergenes in food. And an AI hater will not care, how much AI is in it. Now you backpaddle and argue that you want to have percentages of ai content, rather than mark the game as having ai in it? Do you think that publisher lied about the game having ai in it, and it actually does not? Do you think the publisher lied about all the other games, but not about this?
Also, I have read about the Claire Obscure incident before writing my initial reply. It came up when searching for how the real world deals with the problem you are trying to solve. (With the method I called overengineered). The developers used prototype images made with ai. One of those prototype images was overlooked and made it to release and later was patched out. I would not exactly call that "content". And out of this, they constructed a case how the developer supposedly lied about using ai, and some jury took away an award ... and gave it to the follow up developer that did use ai - but said so. That's kinda a technical disqualification in a competition and not a developer scamming ai haters into buying a game full of ai content.
But pray tell me. How would your solution have dealt with this? Would the developer be removed from the trust network, for "sneaking in" ai "content"? Could your trust network have prevented it? Would you ship the game assets with the watermark header of human generated content and there would be a tech layer of trust between the player and the game on the computer, not rendering the game's images?
And how would this be integrated into Steam? You made a mock up screenshot how you imagine it. If Steam and the developer would adopt your network of trust approach, there would not even be a need for that button. No user interaction needed at all. Steam would automatically check the game against the ai disclosure. And I guess that is something they do by hand and assisted by technology. They do have a process of greenlighting games. You can't just publish anything you want there. They need to approve it. And if the approver sees typical ai content or signs of usage, and you claim you have no ai content, there will be trouble. Steam does has a reputation.
" If your solution blanket blocks almost everything, it will not be very useable in a social media situation."
I did read your whole reply, but this sentence and a few others makes it pretty clear you didn't understand how this technology functions. I'll give you some time to reread the explanation of why it works, but this also makes me think I somehow need to be clearer in the explanatory page. You've shown me which areas need to be explained better, so I appreciate the time taken to comment!
Oh, I think I do understand how this is supposed to work and what you want to do. I am telling you, this will not work, even if your solution would do what you say it would do.
Think about it and try to answer this. Imagine it as a pitch and me being an investor: If your solution cannot prevent and cannot detect a bad actor, but only self heal by removing bad actors*, in what way is your solution any beneft to the regular account scrutiny of any platform - be it publishing or social media.
* and this self heal is not even on a platform level. Bad actors are not removed from the network, but only one line of the graph is blocked. And only by people that previously had a trust connection. And only if they bothered to log in and do that.
I do not think you grasp what a graph is. You basically draw naive daisy chains in your examples. And you rely on isolating the bad nodes, by their immediate neighbors. Completely ignoring the scope and size of a social network. The bad node will have thousands of trust nodes and billions of different paths to you. There is no single point of entry, where only node F would have to block bad actor A, so trusting user U has no longer a connection to A and thus "heal" the network. Your examples show single point of entry for the bad actors, but that is not how networks are. A will have connections to friends F1, F2, F..., F100. And your graph concept needs to have all of them to remove the trust, so no path to U will be available. Also consider shill network nodes S, that could encapsulate A. One could not cut the trust lines to A at all, only by cutting all of the shill nodes S as well - if you would even notice that S are also controlled by A, instead of being naive regular users. So you would need to resort to marking A as an evil node ... and that is just banning accounts with a lot of extra steps.
It gets worse and worse, the more I look into the technical details. That file format is not even needed for your trust graph. And it completely fails for text postings. That trust graph could be stored in the trust network servers. File formats would only be usefull, if you have people outside of the network. Like a seal of approval. Similar to how Microsoft tries to sign executeables.
Calculating paths is costly. Your Kevin Bacon example only works for node sizes of a few hundred. Like known actors. Should your idea be implemented, it needs to be able to calculate paths in a network with billions of nodes. Yes, billions. Facebook has billions of users. Imagine doing path calclulation in such a network.
Your verification by calculating a path is making an O(1) action of checking valid membership to an O(n) action, with n being the number of nodes in your graph.
Anyway, your solution will not work for social media, and it will not work for detecting ai authorship in video games. Those two scenarios are completely different. That you even mention text and claim you would sign the text in a file format shows, how out of touch with reality this concept is for social media.
Oh, and to clarify: I argued under the assumption that your solution would work on a technical level. I have doubts it would, but that is how I argued. I tried imagine a scenario where your solution would be adopted widely. Where people use it and it would do what it says on the tin. And even in that scenario, your concept falls apart when scrutinized by comparison to how the industry deals with the problems you are trying to solve.
Video game developers as a whole are not scammers. And the scams you will encounter are not the type of scam that relies on selling ai works as human made. They do not need to tell lies. So essentially you are trying to solve a non problem. Where scammers do tell lies, is when they claim, that their game is not stolen and does not contain viruses, does not steal your data and does not infect your computer. That is the type of scam you will encounter on Itch. A lot.
