Stories
Slash Boxes
Comments

SoylentNews is people

posted by Fnord666 on Wednesday January 01 2020, @07:58PM   Printer-friendly
from the false-positives dept.

University of Cambridge researchers are hoping to launch technology that blocks online "hate speech" similar to how an antivirus program stops malicious code.

Thanks to researchers at the University of Cambridge, the largest social media companies in the world may soon have the ability to preemptively quarantine content classified by an algorithm as "hate speech"." On October 14, 2019, researcher Stephanie Ullmann and professor Marcus Tomalin published a proposal in the Ethics and Information Technology journal promoting an invention that they claim could accomplish this goal without infringing on individual rights of free speech. Their proposal involves software that uses an algorithm to identify "hate speech" in much the same way an antivirus program detects malware. It would then be up to the viewer of such content to either leave it in quarantine or view it.

The basic premise is that online "hate speech" is as harmful in its way as other forms of harm (physical, emotional, financial...), and social media companies should intercept it before it can do that harm, rather than post-facto by review.

Tomalin's proposal would use a sophisticated algorithm which would evaluate not just the content itself, but also all content posted by the user to determine if a post might be classifiable as "hate speech". If not classified as potential "hate speech", the post occupies the social media feed like any regular post. If the algorithm flags it as possible "hate speech", it will then flag the post as potential hate speech, making it so that readers must opt-in to view the post. A graph from the proposal illustrates this process.

The alert to the reader will identify the type of "hate speech" potentially classified in the content as well as a "Hate O'Meter" to show how offensive the post is likely to be.

The goal of the researchers is to have a working prototype available in early 2020 and, assuming success and subsequent social media company adoptions, intercepting traffic in time for the 2020 elections.


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 5, Insightful) by bzipitidoo on Wednesday January 01 2020, @09:15PM (4 children)

    by bzipitidoo (4388) on Wednesday January 01 2020, @09:15PM (#938364) Journal

    Never mind the use of AI, that's just a distraction, and it will not be any more effective than DRM is. Chatbots who kick participants for saying 4 letter words such as "porn" were nuisances that were never effective at censorship, ridiculously easy to defeat by such things as saying "pr0n" instead, and this Hate O'Meter doesn't sound like it's going to be any better.

    The whole idea of censorship, that it is possible and desirable, and won't backfire, is the problem. Seems every generations has to learn all over again, the hard way, that censorship is just stupid.

    Starting Score:    1  point
    Moderation   +4  
       Insightful=2, Interesting=1, Informative=1, Total=4
    Extra 'Insightful' Modifier   0  
    Karma-Bonus Modifier   +1  

    Total Score:   5  
  • (Score: 0) by Anonymous Coward on Wednesday January 01 2020, @10:08PM (1 child)

    by Anonymous Coward on Wednesday January 01 2020, @10:08PM (#938378)

    You are thinking of filtering, an “AI”. Should learn what you mean from context. But of course that “learning”. Will n the early years seem like misguided opinions.

    • (Score: 1) by khallow on Thursday January 02 2020, @02:06PM

      by khallow (3766) Subscriber Badge on Thursday January 02 2020, @02:06PM (#938602) Journal

      You are thinking of filtering

      Well, that is what this product is, after all. They aren't looking for hate speech so that you can read more of it.

  • (Score: 2) by MostCynical on Wednesday January 01 2020, @10:19PM

    by MostCynical (2589) on Wednesday January 01 2020, @10:19PM (#938381) Journal

    if all it did was add mod points/flag score/whatever and allowed the user to pick a threshold, it *could* work.. but anything that requires the user to *do* something (actively changing threshold/'playing' with settings) will fail - or FB, twitter and other "social media" comments-as-opportunity-to-piss-others-off, sprout slogans or just be dumb... would be more like SN.. and likely, not as 'profitable'..

    --
    "I guess once you start doubting, there's no end to it." -Batou, Ghost in the Shell: Stand Alone Complex
  • (Score: 2) by barbara hudson on Thursday January 02 2020, @05:34PM

    by barbara hudson (6443) <barbara.Jane.hudson@icloud.com> on Thursday January 02 2020, @05:34PM (#938697) Journal
    "Scunthorpe" springs to mind. Getting banned because you live there is pretty stupid.
    --
    SoylentNews is social media. Says so right in the slogan. Soylentnews is people, not tech.