411–420 of 517 posts
Nothing of use here. As per the usual MO of tech companies they throw the responsibility back on the user. Sounds like yet another bullshit clause that they can invoke when they want to cancel you.
I think it's smart to start trying things here. This has infinite flaws with it, but from a business and learnings standpoint it's a step toward the right direction. Over time we're going to both learn and decide what is and isn't important to designate as "AI" - Google's approach here at least breaks this into rules of what "AI" things are important to label: • Makes a real person appear to say or do something they…
> Alters footage of a real event or place I wonder if this will make all forms of surveillance, video or otherwise, inadmissible in court in the near future. It doesn’t seem like much of a stretch for a lawyer to make an argument for reasonable doubt, with any electronic media now.
No, it won't. Just as it does now, video evidence (like any other evidence that isn't testimony) will need to be supported by associated evidence (including, ultimately, testimony) as to its provenance.
> It doesn’t seem like much of a stretch for a lawyer to make an argument for reasonable doubt,
“Beyond a reasonable doubt” is only the standard for criminal convictions, and even then is based on the totality of evidence tending support or refute guilt, its not a standard each individual piece of evidence must clear for admissibility.
>in the future we’ll look at enforcement measures for creators who consistently choose not to disclose this information. Nothing of use here. As per the usual MO of tech companies they throw the responsibility back on the user. Sounds like yet another bullshit clause that they can invoke when they want to cancel you.
I think it's smart to start trying things here. This has infinite flaws with it, but from a business and learnings standpoint it's a step toward the right direction. Over time we're going to both learn and decide what is and isn't important to designate as "AI" - Google's approach here at least breaks this into rules of what "AI" things are important to label: • Makes a real person appear to say or do something they…
Perhaps AI classification is the mirror opposite to porn, using the test: "I'll know it when I don't see it", ie, if an average user would mistake AI generated content for reality, it should be clearly labeled as AI. But how do we enforce this? Does such enforcement scale? What about malicious actors?
We could conceivably use good AI to spot the bad AI, an endless AI cat and AI mouse game. Without strong AI regulation and norms a large portion of the internet will devolve into AI responding to AI generated content, seems like a gigantic waste of resources and the internet's potential.
Earlier quoted context omitted.
Given the regular stories posted on HN about folks who've had some aspect of their social or other media canceled by any some SaaS company, are these companies having many (legal) qualms as it is about canceling people without providing a good reason for it? Would be nice if they did, though...
At the very least it wouldn't be bad for PR if Google bans someone for specifically breaking a clear TOS.
Earlier quoted context omitted.
> * Voice cloning someone else’s voice to use it for voiceover This is interesting because I was considering cloning my own voice as a way to record things without the inevitable hesitations, ums, errs, and stumbling over my words. By this standard I am allowed to do so. But then I thought what does it even mean "someone else's" when multiple people can make a video, if my wife and I make a video together can we not…
> ...to her my voice is someone else. To her, your voice is your voice not someone else's voice. If you share a Youtube account with your wife, "someone else" means someone other than you or your wife. The more interesting and troubling point is your use of "synthetic you" to make the real you sound better!
Earlier quoted context omitted.
There are many cases where such content is perfectly fine. After all, YouTube doesn't claim to be a place devoted to non-fiction only. The first one is an especially common thing in fiction.
Does that mean movies clips will need to be labeled?
I’m reminded of how banks require people to fill out forms explaining what they’re doing, where it’s expected that criminals will lie, but this is an easy thing to prosecute later after they’re caught. Could a similar argument be applied here? It doesn’t seem like there is much in the way of consequences for lying to Google. But I suppose they have other ways of checking for it, and catching someone lying is a signal…
The other thing is that they don’t necessarily want to ban all of this content. For example a video demonstrating how AI can be used to create misinformation and showing examples, would be fairly clearly “morally” ok. The policy being that you have to declare it allows for this sort of content to live on the platform, but allows you to filter it out in certain contexts where it may be inappropriate (searches for election coverage?) and allows you to badge it for users (like Covid information tags).
I think it's smart to start trying things here. This has infinite flaws with it, but from a business and learnings standpoint it's a step toward the right direction. Over time we're going to both learn and decide what is and isn't important to designate as "AI" - Google's approach here at least breaks this into rules of what "AI" things are important to label: • Makes a real person appear to say or do something they…
One black hat thing I'm curious about though is whether or not this tag can be weaponized. If I upload a real event and tag it as AI, will it reduce user trust that the real event ever happened?