A new digital watermarking technique for MIT CSAIL seeks to prevent unauthorized image edits by malicious AI…
Lol what a load of shit. Just let me set the whole picture to blur 1-3% to blend in those pixels… Aaannnddd… Now your face is superimposed onto a naked lady.
Really wanna see how it handles the standard Photoshop touch ups. It’s not like the news media has never altered photos to solicit a skewed perception.
The big problem is AI will (eventually) “see” things as a human does so even in the case that these MIT researchers are able to insert nearly invisible artifacts that fool AI into thinking the edges are different than they actually are, a sufficiently large training set will allow the AI to see that the color borders are more important than artifact borders…
Which will allow AI to bypass this type of watermarking.
I really hate the label AI. They’re data models, not intelligence - artificial or otherwise. It’s PAI. Pseudo Artificial Intelligence, which we’ve had since the 80s.
The thing is that these data models are, in the end, fed to algorithms to provide output. That being the case it’s a mathematical certainty that it can be reversed and thus, shown to be from such an algorithm. Watermark or not, if an algorithm makes a result, then you can deduce the algorithm from a given set of it’s results.
It wouldn’t be able to meaningfully distinguish 4’33" from silence though. Nor could it determine a flat white image wasn’t made by an algorithm.
I think what we’re really demonstrating in all this is just exactly how algorithmically human beings think already. Something psychology has been talking about for a longer time still.
It wouldn’t be able to meaningfully distinguish 4’33" from silence though.
Nor could a human though, no? There’s obviously a lot of metadata about 4’33" that makes it what it is - namely that it is a published work that is performed - but an actual recording of it is silence, so I’m not really sure what this apparent limitation that you’re talking about really is.
Edit: and an AI could observe and analyze that metadata just as much as a human could, provided it has access to it.
You’re following me exactly, just not seeing what I’m pointing at.
I agree, a human can’t meaningfully distinguish between a flat white picture made by a human (with say, MSPaint) and one made by an “AI” with a data model that includes the color Flat White. Similarly there’s no meaningful distinction to be made between 4’33" as performed by an algorithm vs one performed by a master pianist - humans can’t do that and neither can a machine.
We’ve called certain kinds of entertainment “formulaic” - well that wasn’t inaccurate. It was. It is. We are. We are algorithmic. And just like in decades past when scientists put forth the idea that our emotions are just the combination of biology and chemistry, there will be serious existential pushback from certain sectors of humanity. Because it belittles the idea of what it is to be human and relegates us back to simple animals that can be trained. The reality is we are just that. And we keep proving it.
We’ve been seeing this problem framed as one facing teachers and educators: How do we know students aren’t cheating and having an LLM writing their term papers? The reality is if they have been and teachers didn’t catch that from the start? The fault isn’t the tool they used. They’re teaching and grading the wrong thing.
Language, like math largely did with the calculator, will be relegated to machines and algorithms because we already did that to ourselves a long time ago. We’re just building the machines to do the same thing for us, and getting the desired results. If I ask you what 237 x 979 is I don’t expect you to math that out in your head, I expect you to probably use a calculator to get that answer. But it’s still important we teach kids how to multiply 237 and 979 together on paper. It’s very simple to do that and avoid the use of computers altogether. It’s basic writing skills after all. Teaching isn’t about producing term papers, what does it matter that LLMs might be used to cheat them then? It’s about educating the students. Our whole focus on the problems of LLMs is just highlighting over and over and over the problems we as society have had for a long long long time, far before anyone knew what an “LLM” was.
Sorry. I rant.
Am I missing something here? The immunized images just look like they’ve been ‘glazed’ which has been a thing for a few months.
Edit: ah I see, the original glaze could only protect against training and this supposedly will make an image protected against img2img diffusion which explains why it’s visual impact is more pronounced.
