back to article Google to annihilate online trolling with ... tra-la-la! Machine! Learning!

Google and Jigsaw, an Alphabet incubee, hope to tackle online trolling with the launch of Perspective: a new online abuse-detecting service that uses machine learning to highlight “toxic comments.” Late last year, a study by the US Center for Innovative Public Health Research found that 72 per cent of Americans over the age of …

  1. ratfox Silver badge
    1. The Man Who Fell To Earth Silver badge
      FAIL

      You’re a stupid idiot!

      Or in AI parlance, "You're data processing is low speed and high error rate."

  2. Anonymous Coward
    Anonymous Coward

    A-I Fuck Yeah - I guess I'll just leave this here:

    http://www.theregister.co.uk/2017/02/22/facebook_ai_fail/

  3. Destroy All Monsters Silver badge
    Trollface

    What a world to live in!

    Before Trump: Artificial Intelligence is Racist.(will white self-flaggelation never end?)

    After Obama: Doesn't matter, we will use it to squash toxic/racist comments!

    I expect this to be deployed in "safe spaces" to administer a rightful tasering to anyone exhibiting more than his share of daily illiberalism.

  4. Rafael #872397
    Trollface

    Can it detect sarcasm?

    I bet this will work just fine then.

    1. Dave 126 Silver badge

      Re: What a world to live in!

      Another tech blog noted that this Google system currently flags all text written in Arabic as 'toxic', so there is clearly work to be done.

      There has also been the issue of webcams face tracking not working with black people ( the inverse of human controlled CCTV street cameras?), and another machine vision system categorising photos of black people as gorillas.

      It would seem that it is all due to the sample sets that these systems are trained on, as well as the people doing the testing.

    2. Dave 126 Silver badge

      Re: What a world to live in!

      > I expect this to be deployed in "safe spaces" to administer a rightful tasering to anyone exhibiting more than his share of daily illiberalism.

      Look, the best argument you can make against safe spaces is to use the intellectual freedom properly, and not just roll out some clichés or attack some stereotypes. If you're going to fight for the right to express well thought out views regardless of offending people, then think well.

      Yes, there are some idiots on the left who are far too easily offended - and trust me, I talk to them firmly but kindly when appropriate. Even John Stewart would agree with you. However, that doesn't mean that there aren't some systemic issues which add up in a raw deal for many people, be them former coalminers or descendants of plantation workers.

      Anyway,enjoy your Friday!

    3. Ledswinger Silver badge

      Re: Can it detect sarcasm?

      It will have its work cut out, given the proportion of people reporting that they have experienced on line harassment.

      Which, if the numbers are true, would suggest that either a very high proportion of internet users are guilty of behaviour that others perceive as on line harassment, or (improbably) the on line harassers are small in number, but both highly productive and effective in their "task". Whilst not denying the more egregious examples of on line harassment, I simply don't believe the numbers.

      I suspect many people simply have very thin skins when they're online, but fail to see that others may see their responses as the same form of personal attack. Feel free to call me a c**t if you disagree, I won't go crying to Google, asking for protection.

      1. Anonymous Coward
        Anonymous Coward

        Re: Can it detect sarcasm?

        "I suspect many people simply have very thin skins when they're online..."

        It's not just online, there seems to be a general trend these days to take offense at just about anything.

        Often the "offense" is on behalf of other people - which is itself offensive as it assumes that person or group are incapable of complaining when they find something to be genuinely offensive.

        1. Mark 85 Silver badge

          Re: Can it detect sarcasm?

          It's not just online, there seems to be a general trend these days to take offense at just about anything.

          I believe you've nailed it. Political correctness running amok and when things are politically correct (for some value of "correct") there will be an outcry of "I'm offended". Meh.....

    4. Baldy50

      Re: What a world to live in!

      Got to be a joke or fake news, right?

      http://www.zerohedge.com/news/2017-02-23/group-french-protesters-call-obama-run-president-give-french-people-hope

  5. disgustedoftunbridgewells Silver badge

    Troll Trace?

  6. rtb61

    Easy AI

    Here is the model for the AI, is it a comment from a recognised marketdroid, yes, pass. It it a comment from an end user, yes, fail.

    Basically straight up corporate censorship of the internet to favour corporations, Alphabet are scum that secretly tried to pervert the US election system by cooking searches in favour of the corporate whore Clinton, truly disgusting stuff.

    Oh and look, adults need to be censored in order to protect the children, yep, uh huh, we believe you Google.

  7. Jim-234

    In the end just another automated censor to keep you politically correct

    I think in the end all these automated systems will really simply come down to automated censorship to make sure that you don't say anything not perfectly in lock step with the current politically correct thinking for whatever it is that day.

    It just wouldn't do to actually challenge people to think for themselves & face facts or other opinions & actually have to defend their intellectual positions.

    Apparently society is becoming a world run by cry babies & whichever group of cry babies screams the loudest & throws the biggest tantrums, demands that only their ideas are correct & only their voices should be heard.

    Facebook, Twitter & Reddit are already well down the road of censoring anything politically incorrect according to their owners.

  8. cd

    Why NYT?

    Why didn't they just use Youtube comments for testing? They couldn't do any harm there, and there are plenty of examples.

  9. Flakk Silver badge

    "You're a stupid idiot!" is toxic? Really?

    Will an lexiconic arms race develop to bypass the AI, kind of like the "insult battle at the theatre" scene from "Cyrano de Bergerac"?

    "Oh no, Sir, you are too simple! You could have said, 'The alleged positive quality of your perspicacity is egregiously mendacious.'"

    If nothing else, it could serve to increase literacy.

  10. C. P. Cosgrove
    Pint

    Goody !

    Oh Goody ! This means that as a moderator on another forum I won't have to worry about trolls, spammers, moving topics to more relevant sections, soothing ruffled feathers, finding people who know what they are talking about to respond to topics, etc., etc..

    I would drink to that except that I don't expect to be still around when I get automated out of a job ! I will let everybody else worry about the logic in that statement !

    Chris Cosgrove

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2019