Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Google The Internet Technology

Google To Start Fact-Checking Google Images (axios.com) 57

In an effort to curb the spread of misinformation on its platform, Google said Monday it would begin to fact-check Google Images search results. From a report: The danger of text-based misinformation could be dwarfed by that posed by misleading, manipulated and outright fake photos and videos, including convincing "deepfakes" generated with the aid of artificial intelligence. The move, a first among major search engines, comes as tech firms continue to wrestle with how to address the misinformation that continue to run rampant on their platforms. Beginning Monday, Google will surface fact-check information in Google Images around the world, a similar effort to what is does in its regular search and news results.
This discussion has been archived. No new comments can be posted.

Google To Start Fact-Checking Google Images

Comments Filter:
  • by moxrespawn ( 6714000 ) on Monday June 22, 2020 @12:50PM (#60213092)

    If someone applies a tint filter to a picture, does that fail a fact-check?

    Related question: What if it was that O.J. Simpson booking photo?

    • I want to know if the picture on the Wikipedia page for âsVaginaâ really belongs to the chick who claims to have written the article

    • "If someone applies a tint filter to a picture, does that fail a fact-check?"

      Sure. I guess even makeup is.

      PS. Or did you mean an orange tint?

      • For those with the right productive synesthesia, you'll be seeing it all unfiltered very shortly.

    • A picture is worth how many words exactly? I suspect that Google search is going to do its "magic" by comparing new pictures with already indexed pictures. Maybe the algorithm(s), given a set of somewhat identical photos, will flag the older pictures as original and the newer ones as the "fakes"? So if you see, say, a photo of Trump toting a bazooka during a White House presscon, it would be quickly flagged as a fake if a photo already exists of the presscon showing him without the bazooka.
    • If someone applies a tint filter to a picture, does that fail a fact-check?

      Related question: What if it was that O.J. Simpson booking photo?

      That's a good example.

      There's a more recent example -- a black man open carrying that CNN bleached out to look white. How would google's fact checking react to that?

      A larger question might be, who is actually doing the fact checking, and how can we guarantee impartiality?

    • by Tablizer ( 95088 )

      If someone applies a tint filter to a picture, does that fail a fact-check?

      T is really purple, the orange thing is Fake Hues!

  • by Anonymous Coward
    At this point Google may as well rename itself Ministry of Truth. The very fact that Google has such a cozy relationship with the US government and especially the US state department is especially worrying. I think I can just see it now; If you want a vision of the future, imagine a Google logo on a human face - forever.
  • Soooo.... no, really, what's real and what's fake?

    (Verbing Hurray! Verbing Hurray!) Does using Photoshop on a picture to insert an object "fake" it? Does removing an object "fake" it? What about changing the tone, removing, changing, or saturating colors? So like you're saying it has to be an actual picture, no tampering, except for maybe cropping the edges? (That's still a modification from the absolute original.)

    What about optical illusions? Back when photography was first starting out, one guy
    • This is likely just about memes, especially political memes and mainly political memes of a certain persuasion.

      • “We’re also training our algorithms, like, if 2016 happened again, would we have, would the outcome be different?” -- Jen Gennai, Google Global Affairs

        • I see what she was trying to say, but that is some bad phrasing on her part. It comes down to challenging things such as Trump's campaign strategist and counselor, Kellyanne Conway, defending Trump and his staff with what she self-described as "alternative facts" [wikipedia.org], a nonsense term that she used to try and redefine the fact that she was lying. Some have the belief that if Trump and his staff's lies were challenged, that the 2016 election would have ended differently. I personally believe it wouldn't have

          • It's always fascinating how these senators - who are in a position to actually write and promote legislation - are going to magically change into people who "stand up to banks and corporate power" if they win the presidency.

          • "Elizabeth Warren is saying that we should break up Google. And like, I love her but she’s very misguided, like that will not make it better will make it worse because now all these smaller companies who don’t have the same resources that we do will be charged with preventing the next $candidate_we_dont_like situation, it’s like a small company cannot do that." -- Jen Gennai

            Where $candidate_we_dont_like happens to be Trump this time, but could be an actual populist trust buster the next.

        • by ka9dgx ( 72702 )

          If Hillary had bothered to show up where it mattered, she Would have Won... no amount of fscking with the bedrock of society is going to make Biden any more attractive. You can't solve social problems with a technical fix.

    • You're probably overthinking this. If they see an image that looks substantially similar to another image, their index has a strong idea of which one is the "original" and which one is modified/tampered. Tampered images leave digital forensic traces where two images are stitched together. Nearly invisible to the naked eye, but very obvious with even basic image analysis. They'll be able to link you to the original image if a faked duplicate pops up.

      So what again are they trying to do here? No, really, I'm asking

      The number of intentionally falsified photos circulati

    • by Tailhook ( 98486 )

      Soooo.... no, really, what's real and what's fake?

      Google employs a large number of experts that know what should and should not be permitted on the internets. That's why they been empowered to selectively edit whatever they wish while also being immune from any liability for whatever they permit. So don't worry your pretty little head about it.

    • This is really a no win scenario for corporations and governments. People are gullible and stupid, especially for things they want to agree with, but a badly informed or misinformed public is actually dangerous and destructive to democracy yet, there is no possible way to fact check every statement, image, and video. Even with complex algorithms and supercomputers, it's impossible, even if the companies and governments want to play fair, which they rarely do. And these statements don't even touch on unhi

  • There's a broader issue with the people who host the content also moderating it themselves, but putting that aside what legal culpability does Google open themselves to by doing this, because at some point you become responsible for what you fail to moderate as much as you are responsible for what you do.

    If you claim to provide this service and fail to detect something that leads to harm (e.g. failing to catch an image with harmful medical advice) how much liability are you inviting upon yourself? Regard
    • by DRJlaw ( 946416 )

      what legal culpability does Google open themselves to by doing this, because at some point you become responsible for what you fail to moderate as much as you are responsible for what you do.

      And what point is that alvinrod? Articulate it... where does one transition from "cannot be liable for not moderating" to "liable for not moderating sufficiently well" to "not liable for moderating sufficiently well"?

      The Communications Decency Act already says [cornell.edu]:
      (1) No provider or user of an interactive computer service

      • I think he was referring to future possibilities of liability. Already the Republicans, and some Democrats, are pushing to revise section 230 of the Communications Decency Act, which is the section of the law that currently shields companies, such as Google, from legal liability for content that users post. If they modify this section, they could create this theoretical 'middle ground' that currently doesn't exist. Right now, you're either covered by section 230, or you're not, and it's usually pretty ob

    • Filtering images works against what they want. They want everything to be searchable. What they care more about is relevance. Original unaltered images have higher relevance and they want to rank those higher in search results. What's weird is that they're not letting this affect the ranking order (from the article). Instead, they're keeping the original ranking order but putting warnings under the images.

      Where this would be useful is linking to the unaltered form. It would work against anyone's int

      • What's weird is that they're not letting this affect the ranking order (from the article). Instead, they're keeping the original ranking order but putting warnings under the images.

        Using someone's example from above: Imagine someone posted a picture of Trump carrying a bazooka at a press conference. If someone asked me had I seen it, I'd go search for "trump bazooka press conference". If that photo was pushed down to page 3 of the results because it was an altered image, they wouldn't be putting the most relevant results first.

        I'm sure someone in Google is researching actual scenarios, and when they would or wouldn't want to down-rank altered photos. Their decision is going to be base

  • Is Google going to fact check memes?

  • I believe google/facebook/twitter would be better served by not filtering anything. Let their users police themselves.

    • Users are gullible and stupid. This has been proven time, and time again, by many public events and clinical studies. People's critical thinking skills suck, especially against falsehoods that they want to agree with.

      • I don't want to resort to an ad hominem attack but re-read your post. "Peoples critical thinking skills suck....". You sound like an ivory tower academic who thinks that everyone else is too stupid to think for themselves. People are free to think whatever they like, especially when it comes to politics. They are free to read and believe what they want. Why should Facebook or Google employees attempt to censor what you or I believe as a falsehood? This is about censorship and nothing else. If you don't agre

    • Google facebook and Twitter have built up quasi monopolies by offering open platforms. In the current stage however they clearly have decided that their own interests are better served by aligning with anyone with the clout to harm them. And that is not the end user.

  • by slash2019 ( 6030504 ) on Monday June 22, 2020 @02:46PM (#60213562)
    This is simply an attempt to selectively censor free speech that does not match Google's agenda.
    • I wouldn't have put it that way, but my first question was, how do we guarantee the impartiality of the people making the decisions?

      • You hit the nail on the head. This year the censorship seems to favor liberal thinking. Next year it might swing conservative. How is a reader supposed to trust (or distrust) anything in print when they know that someone is cherry picking only articles they think is factual. Remember, there was a time when people really believed the earth was flat. Back then would a round-worlder's comments have been censored? Probably.

        • You hit the nail on the head. This year the censorship seems to favor liberal thinking. Next year it might swing conservative.

          Yeah, it's totally random and could swing either way.

          <chuckle>

      • The whole selling point in making censorship acceptable is 'we're only censoring the bad stuff'. Even Stalin had no objection to that approach. But the problem with Stalin was not only that his judgement of what was bad, was bad...
        This is not just about truth. Even when people are wrong there are reasons to allow them to say what they think. It may matter in a democratic manner, because what they think matters anyway regardless of whether it is right, it may matter because then at least you know what they t

    • I like your explanation, hit the nail on the head.

  • Corrected Headline Reads: Google to start Censoring Political Memes that disagree with their employees Marxist views.

  • World full of artificial lips, injected hips, and fake tits, coming from millions of plastic people representing an artificial life of hype and bullshit.

    Of course fake photos are to blame. Fucking Photoshop and their incessant career-creating capabilities...

    • World full of artificial lips, injected hips, and fake tits, coming from millions of plastic people representing an artificial life of hype and bullshit.

      Of course fake photos are to blame. Fucking Photoshop and their incessant career-creating capabilities...

      Sorry to be pedantic, but it's not the tool, it's the use to which the tool is put. Photoshop isn't evil. It does lots of good things. It's the way some people use it that's evil.

  • Don't make us laugh.

    Whose "fact" anyway? You weren't there. So you ALWAYS only have anecdotal evidence.
    So which fallacy shall it be? Argument from authority? Or argument from popularity? And who defines the authority or by what properties do you measure what is popular. The ones you like the most, in your gut? Your gut as a black-eyed conformist yes man, or your gut as paranoid radical no man

    The whole thing is silly half-think.

    Every. Single. Case. of such behavior is just a pushing of one's own agenda. No m

  • Will this include all the "stock photo" crap the media uses to twist stories for more audience?

  • Hey don't worry, your evil overlords are here to keep you safe from bad information! What could go wrong.
  • You can't solve the world's problems through "fact checking". Because the "fact checkers" have political opinions, and they won't consider their own satires and manipulations and biases to be "misinformation".
  • For over five years when you search for images of "white couples" every picture is of black couples or mixed. Will google fact check and correct those results to show only white couples? Is that what google means by "fact" checking?

UNIX is hot. It's more than hot. It's steaming. It's quicksilver lightning with a laserbeam kicker. -- Michael Jay Tucker

Working...