That’s VERY cool, assuming it works as well as they imply in the abstract.
I’d be curious how easily defeated it is, though. How persistent are the changes to the image if they’re effectively unnoticeable to a human eye? Does that mean that doing an anti-aliasing pass on an image will negate the cloak? What about re-rendering the image at a different resolution? How difficult is it to programmatically determine that an image has had this cloaking algorithm applied (and thus either throw out the sample as bad, or do additional processing to cloaked images)?
It’s going to be an interesting arms race I imagine. Current models do appear to be fairly fragile since they don’t have any inherent understanding of what they’re looking at. From the perspective of the model it’s just a matrix of numbers that it compares to other matrices it’s been trained on to see whether it’s similar or not. This is why small changes in the image that aren’t detectable by a human can have a drastic effect on the ability of the model to recognize the image.
Whether processing the image can negate these changes is an interesting question, and if that’s the case then it would be easy enough to just normalize the images before feeding them to the model.
This is absolutely amazing to see. At the same time, an absolute shame we even need such a thing.
sadly, we do live in a dystopia
It sickens me how normies and even very skilled and knowledgeable people in the tech community can just so actively not care and actively hand over their entire life and information. Yet when something happens that affects them directly, they all of a sudden care, for maybe a month then they just forget about it. Such a shame.
I don’t like to be the dude that says “told you so” but I had a feeling things would turn out this way roughly a decade ago. It was super concerning with the massive push for “just learn to code” spearheaded by Google and friends. I knew we were headed down a dangerous road. Their goal was to monopolize the talent under their terms and they’ve achieved that.
A sentiment I heard a lot over the years boils down to not understanding the libre/liberty vs free as in beer concept. A lot of new comers (I guess they’re not very new anymore by now) are still under the impression that free software means free as in they shouldn’t be allowed to make a living off their code.
A point that really irks me is the bastardization of the understanding of open source. chromium is open source but it’s dictated by Google. They can act like their super “open source” friendly but it’s not really libre. And now chromium dominates the browser market share.
Finally the demonization of social/political issues. It make tech geeks squirm but these issues are inseparable from technology. It used to be mostly apathy but in recent years the social climate seems to have fostered active disdain for ethical considerations of their creations. It’s like software is in the dark ages of ethics compared to other engineering disciplines. Your code affects people.
Hyperbolic analogy but imagine when a bridge collapses and the engineers say “meh, I just build it, their fault for using it”. What other engineering discipline has the gall.
I fear for the day when this generation of greybeards has completely passed. They’re not getting any younger and the line of succession seems awfully thin.
deleted by creator
Yeah, it’s kind of amazing to see how little most people think about what happens to their data. When you talk to people about it, they will typically tell you that they don’t want their personal information used without their consent. Yet, for some reason they just fail to connect that with what actually happens to their data. I do think that education is a big part of the solution here.
And it doesn’t help that education systems are literally locking them into the idea that these things are good, or at least not of worry, due to partnering with stuff like Google for Classrooms, and Zoom, etc. Kids grow up using Apple machines, or even Windows, but don’t bat an eye to even considering teaching kids Linux. Especially because it’s literally free, it baffles me. I understand using Photoshop in certain cases for example as these are still prominent in professional fields unfortunately, and will give them knowledge to use them, but they don’t even consider alternatives, or even show that they exist.
I’m not talking about college here, but more specifically high school and under. These are massive development periods for life, where stuff they learn here sticks with them for the rest of their life, and when you start introducing kids to all these programs, not even discuss issues that may be present with privacy or even collaborative efforts (FOSS). I guarantee you if teachers maybe did small privacy related courses, kids would start to catch on and say “Well why are we using Google Classroom?” and etc, and start to question things. Kids are curious, I mean not everyone, but there will be a lot more kids who maybe start to consider things. Ontop of this, when kids start doing something, other kids follow. As long as they are being educated on these subjects properly that is the most important thing, and as of right now they’re not even being educated on these subjects at all.
For sure, it’s really insidious how kinds end up being indoctrinated into using these services as the norm. GNU has been fighting the good fight for a long time, and it’s really unfortunate how marginalized it’s become in recent years. I see the whole permissive license movement as a step backwards from GNU licensing because it allows companies to freeload on open source without having to contribute anything back.
deleted by creator
deleted by creator
deleted by creator
