> Just the smallest variation in places that have been inked twice will reveal the clear text
Sure. But anyone can visually examine this. That means everyone with situational context can directly examine the quality of the redaction.
Contrast that with a digital redation. You have to trust the tool works. Or you have to separate the folks with context from the folks with techical competence. (There is the third option of training everyone in the DoJ how to examine the inner workings of a PDF. That seems wasteful.)
Can they? In principle it could be the difference between RGB 0.0,0.0,0.0 and RGB 0.004,0.0,0.0, that could be very difficult to visually see, but an algorithm could unmask the data with some correlation.
If you do it digitally and then map the material to black-and-white bitmap, then that you can actually virtually examine.
> Contrast that with a digital redation. You have to trust the tool works.
While true, I think the key problem is that the tools used were not made for digital redaction. If they were I would be quite a bit more confident that they would also work properly.
Seems like there could be a product for this domain.. And after some googling, it appears there is.
> While true, I think the key problem is that the tools used were not made for digital redaction. If they were I would be quite a bit more confident that they would also work properly.
Adobe Acrobat's redaction tools regularly feature in this sort of fuck-up, and they are (at least marketed as being) designed for such use
It's probably fine, but certainly better than what's being discussed ITT.
The larger point is that the "usual" redaction involves a tape pen or paint-style ink (dries opaque), IIRC, then photocopy, because the blocked out area is opaque. Scanner is probably no different than photocopy for these purposes.
They can't, if the variations are subtle enough. For example, many people are oblivious to the fact that one can extract audio from objects captured on mute video, due to tiny vibrations.
Analog is the worse option here. Simple screenshot of 100% black bar would be what a smart lazy person would do.
Perhaps an imagemagick pipeline dumping each page out as a png then blanking areas associated with a list of words (a pixel level concordance of the coordinates of all the words having been compiled from a text dump? Hand-waving here).
I'm probably overthinking this one but the various lengths of the redaction bars would provide some information perhaps? So three conspirators with names like Stonk, Hephalump and Pragma-Sasquatch would be sort of easy to distinguish between if the public had a limited list of people who might be involved?
Booked a trip yesterday, without knowing this has happened. ~1h off my usual trip time, which I got accustomed to in the last two decades. It's extremely awesome!
Yes, but in that story, parent only has the word of that Journalist. I personally don't even have that, I only have a post about it.
My deeper point is that it's arguably very difficult to establish a global, socially acceptable lower threshold of trust. Parent's level is, apparently, the word of a famous Journalist in a radio broadcast. For some, the form of a message alone makes the message worthy of trust, and AI will mess with this so much.
Whether you trust the word of the journalist has little relation to the story. The "socially acceptable lower threshold of trust" is not static for all stories; it changes depending on the stakes of the story.
Non-consequential: A photo of a cat with a funny caption. I am likely to trust the caption by default, because the energy of doubting it is not worth the stakes. If the caption is a lie, it does nothing to change my worldview or any actions I will ever take. Nobody's life will be worse off for not having spent an hour debunking an amusing story fabricated over a cat photo.
Trivially consequential: Somebody relates a story about an anonymous, random person peddling misinformation based on photos with false captions on the internet. Whether I believe that specific random person did has no bearing on anything. The factor from the story that might influence your worldview is the knowledge that there are people in the world who are so easily swayed by false captions on photos, and that itself is a trivially verifiable fact, including other people consuming the exact photo and misinformation from the story.
More consequential: Somebody makes an accusation against a world leader. This has the potential to sway opinions of many people, feeding into political decisions and international relations. The stakes are higher. It is therefore prudent not to trust without evidence of the specific accusation at hand. Providence of evidence does also matter; not everything can be concretely proven beyond a shadow of a doubt. We should not trust people blindly, but people who have a history of telling the truth are more credible than people who have a history of lying, which can influence what evidence is sufficient to reach a socially acceptable threshold of trust.
The point about the stakes is a good one. But there is an individiual factor to it. And maybe it's exactly because of the stakes you mention: if you perceive your personal stakes to be low, or might even gain something out of redistributing the message, no matter if fabricated or not, your threshold might be low as well.
> > Trivially consequential: Somebody relates a story about an anonymous, random person peddling misinformation based on photos with false captions on the internet. Whether I believe that specific random person did has no bearing on anything.
> The point about the stakes is a good one. But there is an individiual factor to it.
Indeed. The so called "trivially consequential" depends on whether you're the person being "mis-informationed" about or not. You could be a black man with a white grandchild, and someone could then take a video your wife posted of you playing with your grandchild, and redistribute it calling you a pedophile, causing impact to your life and employment. Those consequences don't seem trivial to the people impacted.
This is a complete and total misrepresentation of what I said. The key point here is that the "accused" in the trivial story is anonymous. They are fungible. Their identity is irrelevant to the story; it is merely an anecdote about the fact that a person like this exists, and people who exhibit the exact same behaviour as them verifiably do exist, so there is nothing to be misinformed about. A tangible accusation against a specific individual is completely different, and obviously is consequential.
Who cares about a single or two Yachts. Ukraine likely made 100 billion USD disappear and there were many people expecting just that. Just like some of the "donated equipment" started showing up on all sorts of black markets once it was shipped to Ukraine. It's just the obviously controlled media in Europe that stopped mentioning Ukraine's corruption issues right after February 2022.
Obviously I can only be a Putin-loving propaganda bot for saying such things.
Everybody is aware the Ukraine has major corruption issues. It is frequently covered in the media and is common knowledge.
I have no doubt however that Europe (and hopefully the wider world) is less worried about that corruption than they are about Russian military aggression. And there will be some level of media focus on that – rightly so, where the focus should be on grinding the Russian kleptostate into dust as quickly and thoroughly as possible.
You're not a propaganda bot; you're just making their lives easier.
If you're happy with your tax euros disappearing in Ukraine, good for you.
I know for a fact via family ties that major newsrooms in Germany received instructions to tune out the corruption angle once the war started. I'm sure it's all nothing though and that Putin will find himself in Poland next year. Of course!
If sending hundreds of billions of tax payer money to a known oligarch run cleptocracy is comparable to some German conservative party affiliate making a couple of millions using shady COVID mask deals is comparable to you, I rest my case.
It's all corruption in the end so who cares, right?
Two things can be true at the same time - we don't want Russia to absorb Ukraine and then further threaten the eastern border of the EU, and we don't want Ukraine to be corrupt.
And in the Ukraine we see that the corruption is uncovered punished, even if it is in the direct circles of the president.
There are problems in uncovering it, but the attempt to get rid of corruption is a big factor in the whole situation and one of the things Russia fears.
For Russia a corrupt system was a lot simpler to influence and Ukraine showing how a partially Russian speaking country, where people moved back and forth, fighting corruption was a threat to the system.
Ukraine was, and is still, one of the most corrupt developed countries in the world. Whether it is slightly more or slightly less corrupt than Russia I do not know. Both are Oligarchic in nature. In my opinion one of the reasons the various peace deals have not succeeded yet is because they fail to acknowledge the Oligarchic nature of both states and that they will both need to continue in that mode going forwards, probably as a frozen conflict or in a system where it is in neither interest to disrupt the balance (because it would end the corruption, pocket-lining, theft, etc). Of course for the ordinary unfortunate Ukrainian, well he/she matters little or not at all to rulers on either side.
Hah, Kim Dotcom is still around? In the 90s he was bragging that he's this super hacker that made millions, his website posted pics of parties, cars, girls, and yachts, and it turned out those were bought/rented using swindled investor money (ironic that he's accusing Zelensky of the same crime). Then he became a sort of hero when the US/NZ governments Team 6-ed his house for the crime of aiding copyright infringement.
Still around and at a huge cost to the New Zealand taxpayer - people used to have some sympathy but public opinion turned against him years ago. His extradition was declared ok, long overdue he was put on a plane and made somebody else's problem.
The people of New Zealand should be mad at the illegal tactics used by the FBI and GCSB. And why should he be extradicted to a country he never visited?
A common pattern you'd find in reliable research papers is that authors tend to understate their findings, which in practice strengthens the impact of their conclusions.
You see, we are here observing a clash in the terminology. Hallucinations in humans is thinking, just not typical. So called "hallucinations" in LLM programs are just noise output, a garbage. This is why using anthropomorphic terms for programs is bad. Just like "thinking" or "reasoning".
I think the answer is somewhere in the middle, not as restrictive as parent, but also not as wide as AI companies want us to believe. My personal opinion is that hallucinations (random noise) are a fundamental building block of what makes human thinking and creativity possible, but we have additional modes of neuroprocessing layered on top of it, which filter and modify the underlying hallucinations in a way so they become directed at a purpose. We see the opposite if the filters fail, in some non-neurotypical individuals, due to a variety of causes. We also make use of tools to optimize that filter function further by externalizing it.
The flip side of this is that fundamentally, I don't see a reason why machines could not get the same filtering capabilities over time by adjusting their architecture.
I have never in my life met a person who hallucinates in the way ChatGPT etc do. If I did, I would probably assume they were deliberately lying, or very unwell.
That's not very competent.
> going analog is foolproof
Absolutely not. There are many way's to f this up. Just the smallest variation in places that have been inked twice will reveal the clear text.
reply