Platforms remove graphic videos, often using artificial intelligence - but footage that may help prosecutions can be taken down without being archived. Meta and YouTube say they aim to balance their duties to bear witness and protect users from harmful content.
But Alan Rusbridger, who sits on Meta's Oversight Board, says the industry has been "overcautious" in its moderation.
The platforms say they do have exemptions for graphic material when it is in the public interest - but when the BBC attempted to upload footage documenting attacks on civilians in Ukraine, it was swiftly deleted. Artificial intelligence (AI) can remove harmful and illegal content at scale. When it comes to moderating violent images from wars, however, machines lack the nuance to identify human rights violations.
Human rights groups say there is an urgent need for social media companies to prevent this information from vanishing. "You can see why they have developed and train their machines to, the moment they see something that looks difficult or traumatic, to take it down," Mr Rusbridger told the BBC. The Meta Oversight Board that he sits on was set up by Mark Zuckerberg and is known as a kind of independent "supreme court" for the company, which owns Facebook and Instagram.
"I think the next question for them is how do we develop the machinery, whether that's human or AI, to then make more reasonable decisions," Mr Rusbridger, a former editor-in-chief of the Guardian, adds.
[...] Human rights campaigners say there is an urgent need for a formal system to gather and safely store deleted content. This would include preserving metadata to help verify the content and prove it hasn't been tampered with.
Ms Van Schaak, the US Ambassador for Global Criminal Justice, says: "We need to create a mechanism whereby that information can be preserved for potential future accountability exercises. Social media platforms should be willing to make arrangements with accountability mechanisms around the world."
(Score: -1, Troll) by Anonymous Coward on Friday June 02 2023, @09:11PM
Let everyone upload their snuff videos to social media, and don't take them down. Deal?
(Score: 4, Informative) by krishnoid on Friday June 02 2023, @09:35PM (15 children)
Human body open on a table: medical information
The same thing, but on a road: violent image from war
I'd think the AI would consider any violent images from war as evidence of human rights violations. Some people would too.
(Score: 4, Informative) by RS3 on Friday June 02 2023, @11:21PM (13 children)
Throughout history human societies and governments have based rules, laws, punishments, etc., on some set of ethics, rights and wrongs, etc. The USA's Declaration of Independence contains (one of my favorite lines) "We hold these truths to be self-evident...", but who sets those truths? Today's society, seemingly more than ever, seems to be struggling with what's right vs. wrong.
So I keep wondering: what rules, guidelines, etc., do AI use to evaluate what's good, bad, or indifferent?
(Score: 3, Insightful) by Immerman on Saturday June 03 2023, @12:03AM (12 children)
>but who sets those truths?
The people who said the words - I know the language is a little flowery, but they literally just said "these things are true because we believe them" The Crown and nobility most certainly disagreed (as do most people to some degree, even today)
As for AIs? They'll use whatever guidelines we've trained into them - guidelines that almost certainly only vaguely align with what we were *trying* to train into them, and absolutely should not be trusted to remain remotely consistent with our intent anywhere outside the limited domain that was thoroughly sampled by their training data.
(Score: 2) by RS3 on Saturday June 03 2023, @12:13AM (10 children)
Thank you, that's what I've been concluding. I'll augment: if the AI is quite independent, what's to stop it from deciding on what it thinks should be right or wrong? Criminals, even the worst of the worst, usually have justification for what they've done. So again, what's to stop an AI from concluding its own set of rules, regardless of what it was originally programmed with?
(To be fair and give due credit, in the movie "2000: A Space Odyssey", the computer HAL 9000 ends up in conflicting priorities, and, well, I don't want to give it away in case someone hasn't seen it but may wish to. It's better clarified in the followup movie "2010: Odyssey Two").
(Score: 2) by Immerman on Saturday June 03 2023, @04:21AM (9 children)
Mostly the total lack of consciousness, awareness, etc., without which it can't really "decide" anything. It's a deterministic system that can only act in the way we've trained it to act, even if we don't actually know what that is.
For a hypothetical "Realized" AI that *is* aware? I'd say the answer would be exactly the same as for a human - absolutely nothing.
(Score: 2) by RS3 on Saturday June 03 2023, @05:02AM (8 children)
I may be unclear of the current usage definition of AI. I was expecting, and from reading some articles about AI, that it kind of "thinks" and is somewhat self-aware. But I'm no expert on AI- too many other things to do in life.
(Score: 2) by mhajicek on Saturday June 03 2023, @06:02AM
Current AIs are deterministic digital computer programs, just far too complicated for a human to analyze.
The spacelike surfaces of time foliations can have a cusp at the surface of discontinuity. - P. Hajicek
(Score: 2) by mhajicek on Saturday June 03 2023, @06:03AM (6 children)
Though one can argue that human minds are also deterministic.
The spacelike surfaces of time foliations can have a cusp at the surface of discontinuity. - P. Hajicek
(Score: 2) by RS3 on Saturday June 03 2023, @07:12AM (5 children)
I started but deleted a longer response. Does it come down to randomness? In other words, not on some absolute scale, but can humans be random enough, at random times, such that it would be effectively impossible to determine what would come next?
If you think it's possible, I'll introduce you to one of my past girlfriends. To be fair, she's an amazing artist (and singer / songwriter / musician / performer).
(Score: 2) by mhajicek on Saturday June 03 2023, @04:04PM (1 child)
I believe in an entirely deterministic universe. Even quantum effects, which appear random to us; while we have no way to predict them, I believe they were always going to happen exactly that way.
The spacelike surfaces of time foliations can have a cusp at the surface of discontinuity. - P. Hajicek
(Score: 3, Insightful) by RS3 on Saturday June 03 2023, @04:17PM
I'll give you that, given an essentially infinite amount of data of infinitesimal resolution. As much as I love philosophy, I'm an engineer, so useful / practical is high priority for me, and I like to think in relative / contextual ways. It's getting to where I don't know what to believe any more, maybe apply skepticism everywhere. That said, many developers of AI say they don't even know what's going on inside the processing. Meaning, sure, theoretically you could wire in test points and monitor every electrical signal, but, well, that would be pretty impossible in practice. But always good to keep the biggest picture in mind, and strive for the best we can do, but of course accept the practical limits.
(Score: 3, Insightful) by Immerman on Saturday June 03 2023, @06:11PM (2 children)
There's kind of two different answers depending on exactly what you're asking.
Does non-determinism come down to randomness? Pretty much. Free will is another possibility, though there's not really any conclusive evidence that such a thing exists.
Does awareness, free will, etc. come down to randomness? Assuming it exists at all of course, it does seem likely that randomness plays some role. Though it's likely not sufficient on its own or any dice throw or quantum system would be conscious.
Of course panpsychism takes exactly that position, was quite widely accepted in the ancient world, and is regaining acceptance in recent years as an explanation for how consciousness can exist at all. There's really only three options: It's a property of a non-physical "soul", for which we've never found any evidence. It somehow arises from non-conscious systems, as has been the prevailing theory for at least a century or so without having found any supporting evidence. Or it's an innate property of everything, and higher consciousness such as ours is an emergent property of organized interaction of countless lesser consciousnesses - not entirely unlike how a mob behaves as a collective consciousness that largely subsumes the normal individual consciousness of the people within it (I understand the apparent loss of individuality and free will while in a mob can be extremely unsettling after the fact, especially to people who are unusually self-aware or self-controlled)
Can humans be "random" enough to be unpredictable? Absolutely, on an individual scale, at least in practice. The larger the group though, the easier we become to predict.
Can humans *actually* behave randomly? Unknown. It's not certain that *anything* can - it may be that the universe is completely deterministic, but it's not possible for us to know enough detail to predict it. Not just we don't have the technology to record sufficient detail - the Heisenberg uncertainty principle has been extremely well tested and established that there is a maximum amount of information that can be measured about (the physical aspects of) a quantum system, and that their are no hidden proper - e.g. the more accurately you measure something's position, the less you know about its velocity, and vice versa. If you somehow perfectly measured the velocity of a particle, it's position would become so uncertain that it could be anywhere in the universe.
There are possibly ways around that in some interpretations of quantum mechanics how the universe actually works - such as Bohmian Mechanics/Pilot Wave Theory or Super-Determinism, which postulate that quantum systems are actually deterministic and governed by non-local hidden variables (e.g. some sort of unified universe-spanning quantum wavefunction). As yet there's little compelling evidence to support (or dismiss) such interpretations, and the theories currently have no answer to the much more sophisticated and accurate Quantum Field Theory that emerged from the Copenhagen Interpretation... but in fairness there've been orders of magnitude fewer researchers pursuing such interpretations, so that's to be expected.
And of course none of that makes it any easier to *predict* what will happen, since even if we could eventually measure the universe-spanning unified wavefunction directly, there's still no practical way to measure it at at all points in the universe in order to fully characterize it so that you can predict the behavior of the atom in front of you... but at least in theory some sort of omniscient god could do so, and would then be able to accurately predict exactly how the universe will progress (at least assuming they also have enough computing power to do so).
(Score: 2) by RS3 on Saturday June 03 2023, @07:06PM (1 child)
You've elucidated my thoughts exactly. Or something like that. But seriously, thank you for expanding on what I was saying, and your mention of Heisenberg leads me to think that, within our current knowledge and context, we can't know all possible outcomes of something even if we could track all inputs and functions of the system being observed and predicted.
But we know there's much we still don't know. Again, in the case of AI, there's much more going on inside than what the developers can know, so in a limited way it is its own entity.
How to define "sentient" gets messy and is beyond my realm of curiosity, but it seems like we're not far off, if someone hasn't done it already. I've heard one measure is "self awareness"...
I'm curious what your profession is, and are you a teacher / professor, or consider going into that field?
(Score: 3, Interesting) by Immerman on Saturday June 03 2023, @08:51PM
Yeah, under the current broadly accepted understanding of quantum mechanics, perfect prediction of the future is theoretically impossible.
We *can* theoretically know all possible paths, outcomes, and their probabilities (Feynman diagrams leverage that at the particle level), it's just which one will actually end up happening that's impossible to predict.
And yeah, there's definitely some big holes in our understanding. Just on the level of fundamental physics we've got a couple great big theoretical conflicts between QM and Relativity, and a whole bunch of unexplained phenomena. Get into stuff like minds or ecosystem webs where the level of complexity appears to be way beyond our ability to even think about that much information? I wouldn't be surprised if there are some such mysteries we never solve.
I don't know that I really have a particular profession. I tend to work in programming or technology, but have yet to find anything I could see myself still doing in ten years (well, I always program as either work or hobby, but to always do it as work?). I'm a lifelong student of whatever catches my fancy, with degrees in computer science, mathematics, and *almost* engineering ( I wasn't willing to stick around waiting for a few rarely-offered classes. )
I love and generally excel at teaching in a conversational setting, which is a big part of why I make posts like the above, but while I did teach at local colleges and universities for a few years I found it to be grueling work that forever left me feeling that I should be doing more for my students, and I'm not sure I was actually all that great at teaching in that setting anyway. Seems like it might be more satisfying at the "tenured at a big-university" level where your job is more to paint the big picture and inspire the students to do the legwork for themselves, with grad students to do all the grading, etc. But that's a lofty goal (and a lot more school) for a "might be".
(Score: 3, Insightful) by legont on Saturday June 03 2023, @04:26AM
Its easy to brainwash humans because they are social and vast majority of them are stupid. One targets idiots and then even the smartest either join the herd or go nuts.
AIs? I don't think so. One could make them easy targets for manipulation, but it defeats the purpose of AIs.
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 2) by driverless on Saturday June 03 2023, @09:48AM
There's a different type of context:
In this case it's both valuable evidence of war crimes and a snuff movie.
Yes, I know the latter is an urban legend, but footage of an actual person being killed live on camera is what a snuff movie is supposed to be. The content hosters are in a rather awkward position here, do you want to be the person explaining to Focus on the Family why your organisation is distributing snuff movies to children?
(Score: 4, Insightful) by VLM on Friday June 02 2023, @09:39PM (3 children)
Someone's gotta say it:
"Social media censorship is a human rights violation"
(Score: 0) by Anonymous Coward on Saturday June 03 2023, @12:10AM
(Score: 2) by Azuma Hazuki on Saturday June 03 2023, @09:43PM (1 child)
No it isn't. It can be used as part of a human-rights-violations toolkit, but in and of itself, it's not. And don't you fucking dare compare getting your bad takes handed back to you freshly flame-grilled to a platform deliberately removing evidence of war crimes. You are not a persecuted minority victim of jack and/or squat.
I am "that girl" your mother warned you about...
(Score: 2) by VLM on Sunday June 04 2023, @01:45PM
Or what?
If all your argument has going for it is aggression to the level of poor taste mixed with sophistry, ever consider that might be evidence you're wrong?
(Score: 2) by SomeGuy on Saturday June 03 2023, @12:35AM (1 child)
Oh, I don't know, how about not relying on social media to present or preserve evidence?
Because, of course they are going to filter stuff on their platform to meet whatever social or political ideals that the day holds.
Take a video of a crime, and you should send it directly to the police. Post it on twitter or facefuck? Or even Youtube? Thats stupid. Crime is bad, so CENSORED. (But first an unskippable 2 minute advertisement!!!!)
You want human rights managed by a machine? Really?
(Score: 0) by Anonymous Coward on Saturday June 03 2023, @01:01AM
> send it directly to the police
Makes sense, but how do I do that? Somehow I don't think the cops are set up to deal with this kind of evidence (archiving, backing up, indexing, etc), and oh by the way, protect the sources too.
(Score: 2) by Rosco P. Coltrane on Saturday June 03 2023, @12:54AM (2 children)
I imagine if someone is being tried for war crime, the evidence isn't a fucking URL.
Everybody who's ever downloaded a pirated movie or music album knows anything on the internet is temporary. It's liable to disappear any minute because sites die, people yank what they put up online, the RIAA/MPAA go on a rampage or our new privately-run public spaces called "soclai media" are heavily censored. If I was interested in gathering war crime evidence on the internet, the first thing I'd do is save a copy.
As to whether it's important to disseminate the knowledge and details of instances of war crime, if the material is cringy or NSFW and it's not accepted on the usual sissy social media sites, then it needs to be posted elsewhere.
(Score: 5, Insightful) by ElizabethGreene on Saturday June 03 2023, @04:09AM
Why would you imagine that? Do you think CNN had cameras in Tikrit in 2014 when ISIS assembly-line massacred a thousand men by walking them to a pier and shooting them one-after-another? No. ISIS recorded it, ISIS published it to their supporters, and now that footage has been taken down so now only the bad guys and a few people thoughtful enough to scrape copies have it.
Contrary to popular belief, things on the internet are not there forever, particularly when those things are gross TOS violations for major platforms.
(Score: 3, Insightful) by legont on Saturday June 03 2023, @04:36AM
Everything on internet is archived. Every single bit. Encrypted stuff too for future decryption which one day will undeniably happen. That is the goal.
It's just folks who do it don't advertise themselves.
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 0) by Anonymous Coward on Saturday June 03 2023, @06:26AM
News at 11.