back to article Oi, clickbait cop bot, jam this in your neural net: Hot new AI threatens to DESTROY web journos

Artificial intelligent software has been trained to detect and flag up clickbait headlines. And here at El Reg we say thank God Larry Wall for that. What the internet needs right now is software to highlight and expunge dodgy article titles about space alien immigrants, faked moon landings, and the like. Machine-learning …

  1. Anonymous Coward
    Anonymous Coward

    Click

    Oh no! I fell for the "Hot new AI threatens to DESTROY web journos" clickbait!

    1. big_D Silver badge
      Headmaster

      Re: Click

      The sooner this is implemented, the better! :-D

      I'll still use the RSS feed to manually decide what to read... If my RSS reader implements this, I'll hardly get shown anything, as I only have El Reg and Thurrott in my feed...

    2. JimboSmith Silver badge

      Re: Click

      I like the

      You won't believe this banking/shopping/smartphone etc.trick actually works!

      because if I won't believe it why should I waste data looking at it? Besides I've blocked the most frequent purveyors of these things in noscript, so I'm not very likely to click on them anyway.

  2. Oengus Silver badge
    Joke

    What is clickbait?

    Just about every item on page one of any search engine's results (especially Google).

    1. Ledswinger Silver badge

      Re: What is clickbait?

      There's a much simpler definition that the researchers could have used, and probably does far better than 80% accuracy: Any headline containing an unnecessarily capitalised word.

      Get double clickbait points if that word is "shocking" or "banned".

      1. Lyndon Hills 1

        Re: What is clickbait?

        Or words ending in! ??

  3. Mycho Silver badge

    Can we start with youtube videos that are speech synthesizers reading news articles in front of a screengrab of something vaguely related to the story in question?\

    I was trying to find a politician's recent speech on youtube and gave up after about fifty of those.

  4. Anonymous Coward
    Anonymous Coward

    what exactly is a clickbait headline? It's a tough question

    tough question, soft answer: it's a headline that baits clickers to click on a c...

    1. I ain't Spartacus Gold badge

      Re: what exactly is a clickbait headline? It's a tough question

      Maybe clickbait has just changed?

      In the good old days, the art of headline writing was to get you interested in the article so that you'd read it. And it didn't matter anyway, as you'd already bought the newspaper, so they'd already been paid.- so what matter if they just used it to line the cat's litter tray?

      So your headlines were more about generating a house style and could develop into humour/entertainment. Or be painfully serious and earnest, if that was the house style.

      But then things changed. So I now see clickbait as denying information to the reader. I "read" the Guardian website every day. But given how low their standards have dropped with the new editor, I might only click on one or two articles. Also if the headline tells you the story, such as "murder suspect arrested" you don't need to read it unless you've more than a passing interest. So I'm still scanning them for a bit of an update on news, but they're not making much money. Incidentally the Telegraph have now got so poor that I don't even bother to scan their headlines in hopes of the odd good piece.

      Hence the headline is now about telling you as little about the story as it can get away with, while still getting you to click.

      There's an online only site called The Canary. Very lefty and so hyper-partisan that I'm not even sure it qualifies as news anymore. I've looked at it every so often, out of interest in the Corbyn phenomenon. And they don't have a single headline that tells you what the story is. You can't get any news from the front page, other than learning which politicians they hate. The only way to know if the article is worth reading, is to click on it - because the headlines are so deliberately obscure.

      This turns out to be because, in a not very socialist manner, the writers of the articles are paid purely by the number of clicks they get. So the "better" they can make their headline, the more cash they'll get. And they don't put bylines on there, so you can't even learn to avoid the worst people. Although having read it a bit, I'm not sure any of the writers are any good.

      Then I'd argue there's the subset of this of trolling your readers for a reaction. Daily Mail / Guardian style. Get a nice bit of outrage going, and then watch the comments section light up.

      The Register are still confident enough that we'll click on enough articles, that they haven't resorted to this. Though I suppose they do have the clickbait of headline puns that are so good/bad that I want to click on a piece I'm otherwise totally uninterested in, just so I can congratulate the subbies in the comments.

      1. Robert Carnegie Silver badge

        Re: what exactly is a clickbait headline? It's a tough question

        I count Register puns - not to mention the rhyming headlines - as a reason not to read. If your story doesn't hold your own attention......

        1. I ain't Spartacus Gold badge

          Re: what exactly is a clickbait headline? It's a tough question

          Robert Carnegie,

          It's just house style. Trying to keep up the irreverent atmosphere. After all, Storage vendor brings out new product that's 5% better than last years', ain't exactly thrilling and wonderful It's dull. But if your job involves IT storage, then you might want to know it. So El Reg's style is to mix industry news and comment with a bit of purely silly stuff and some interesting science and a bit of vaguely tech related politics - to make itself more attractive to its readers.

          The silly headlines appear when the subbies have the time and inspiration to do so. And long may it continue. Though I confess to being bored of the SuperCali type ones. But I wouldn't wish that personal preference to come between a Reg subbie and their right to twist that particular trope just a little bit more.

  5. Anonymous Coward
    Anonymous Coward

    "Unfortunately, the paper doesn’t share too many details on what patterns the model picked up, such as which combination of words were more likely to be identified as clickbait"

    so the paper itself was clickbaity?

    1. Michael Wojcik Silver badge

      Heh. In all seriousness, though, it's generally quite difficult to explain what features have been developed by unsupervised training of a convolutional neural network or LSTM network (which is a type of recurrent neural network that has a relatively complex state).

      It's quite likely that there are various "combinations of words" which result in a high clickbait score in some headlines and a low score in others.

  6. Pete 2 Silver badge

    Something bad happened! Read more here.

    > The trouble is, what exactly is a clickbait headline?

    It is one that imparts no information. If you want examples, just look at the Daily Express, The Guardian or any other trashy online newspaper.

    They typically have headlines that ask a question that almost always complies with Betteridge's law (i.e. the answer is "no"). Or that feed on fear, or that bait a reader to continue reading an article.

    The problem with having AI write news articles, or to detect clickbait, is that sooner or later those same AIs will be trained to write irrelevant, clickbait, article themselves. Though we can probably take solace that they will be better at it than people, so all the worthless news website employees will still get sacked. Though we still won't get any better quality written news.

    1. amanfromMars 1 Silver badge

      Re: Something bad happened! Read more here.

      The problem with having AI write news articles, or to detect clickbait, is that sooner or later those same AIs will be trained to write irrelevant, clickbait, article themselves. ... Pete 2

      Are we to assume and/or presume that AI presently then is trained to write relevant, although sometimes in recognition of the need for a little more effective security, suitably irreverent and difficult to believe articles/commentary/Advanced Basic Sublime Human Programming?

      Or are you led to believe to not expect Virtual Machines and/or AIBots and Bodies to be so capable and enabling ...... because of the Madness and Mayhem, CHAOS and Disruption and even Mass Destruction that humans would wrought on the news?

    2. katrinab Silver badge

      Re: Something bad happened! Read more here.

      Or for better examples, look at Outbrain and Taboola.

      1. Anonymous Coward
        Anonymous Coward

        Re: Something bad happened! Read more here.

        I am pretty certain that most of the articles on the Daily Fail are written by a machine, particularly the ones about female 'celebs' in their bikinis which use the exact same wording every time wether its negative or positive .

        Er, how do I know... er... my girlfriend reads the articles and I see the pictures by accident, honest guv!

        1. katrinab Silver badge

          Re: Something bad happened! Read more here.

          Sorry, being in a relationship with a Daily Mail reader doesn't get you off the hook.

  7. m0rt Silver badge
  8. JimC Silver badge

    Therein lies the problem - its not the headline..

    A headline is designed and intended to be clickbait. What makes it the undesirable side of clickbait is not the headline itself, but the content that it leads to. So, I submit, what is needed is Ais that can determine how worthwhile the content is. Some factors might include:

    - More than 300 words of real content per page:

    - Less than 3 pages total unless very high word counts

    - Images that are not reproduces endlessly elsewhere

    - Thumbnail images that are contained in the first page

    - unique text

    and I'm sure we could add many more...

    1. techmind

      Re: Therein lies the problem - its not the headline..

      Destination page...

      - More than 50% ads by surface area

      - Average reading-age score of any textual content less than 10-years-old

  9. Crisp Silver badge

    This hot industry journalist has just finished banging out a headline

    And you wont believe what they do next!

  10. Giovani Tapini

    I think I can define ClickBait as

    Taboola.

    Their whole business is based on farming clicks to ad-loaded pages with multiple click through to access limited and often made up content.

    Usually the headlines are accompanied by unlikely looking photoshopped images to trap the unwary.

    I am sure there are similar publishers but this one stands out as providing nothing but adds and naff content. AI should train on them...

    1. I ain't Spartacus Gold badge

      Re: I think I can define ClickBait as

      Outbrain is the other one that springs to mind.

      What shocks me is that reputable publications allow this shit on their pages. I can't believe the pennies they get paid for those clicks are worth the damage it does to their reputation.

      1. Jamie Jones Silver badge

        Re: I think I can define ClickBait as

        Ugh. Yes. Outbrain and tombola - sites I completely block in my firewall.

        Total dregs of the internet.

        1. onefang Silver badge

          Re: I think I can define ClickBait as

          "Outbrain and tombola - sites I completely block in my firewall."

          Until now I have never heard of those two sites. By the sound of it, hopefully I'll never hear about them again.

          1. I ain't Spartacus Gold badge

            Re: I think I can define ClickBait as

            onefang,

            It's the shit that turns up with lots of pictures at the bottom of articles on better sites. Usually very focused on clickbait headlines about celebs and bullshit medical/diet info.

            I barely register it, because 20 years of exposure to the internet has caused me to go ad-blind. My brain has leared the bits of the pages not to look at, and now doesn't see them unless I'm looking for them specifically. Also I try to avoid clickbait headlines as a matter of principal.

            I picked up much of my knowledge about their shit from the brilliant Dave Gorman's 'Modern Life is Goodish'. Apparently one of their tricks is the clickbait headline, "You'll never guess which celebrity has got [insert horrible disease]." Illustrated with a picture of someone properly famous like George Clooney. Only when you click on the article it's some nobody from "reality" TV. I think he even showed versions of this where it was "You'll never guess who just died!" with a celeb photo and then the article about someone entirely different.

            Anyway they're the pond scum of the internet - and sites who use their shit should be ashamed. They flog dodgy diet pills and crackpot, unscientific diets - so I suppose they're a perfect fit for the Daily Mail...

            1. Jamie Jones Silver badge
              Thumb Up

              Re: I think I can define ClickBait as

              @Not Sparticus, you make a great point about these sites showing up on supposedly "proper" sites - I think that's what annoyed me most about them - my guard was down somewhat.

              And yeah, I suppose they think the pennies they earn are worth the reputation knock.... It certainly makes me think twice of a site that uses such click farms (though of course, I don't usually see them any more)

          2. Jamie Jones Silver badge

            Re: I think I can define ClickBait as

            "Outbrain and tombola - sites I completely block in my firewall."

            Until now I have never heard of those two sites. By the sound of it, hopefully I'll never hear about them again.

            Arggh, despite Giovani giving the correct name, I cocked it up in my reply.

            As he says, it's taboola not tombola

            For completeness, here is my personally compiled list of similar clickbait sites. Of course, check them out before blocking them - some may be dead now, or may have cleaned up their act. More importantly, never trust random-person-on-internet-especially-when-hes-a-self-described-welsh-git!

            # --------------------------------------------------------------------------------------------------

            # Kill these zones belonging to either deceptive ad companies, or crappy tacky 'content based'

            # click-bait ad-servers, which generally link to a page that links to the link (along with

            # others). They usually have a misleading sensationalist headline too. Kill them with fire!

            .adhitz.com. domain

            .adsmarket.com. domain

            .adnxs.com. domain

            .content.ad. domain

            .content-ad.net. domain

            .gravity.com. domain

            .mgid.com. domain

            .outbrain.com. domain

            .zizu.xyz. domain .steepto.com. domain

            .taboola.com. domain .tribalfusion.com. domain

            .zergnet.com. domain

            .revcontent.com. domain # This one even uses thumbnails of people irrelevant to the story! # --------------------------------------------------------------------------------------------------

            #

  11. Paratrooping Parrot
    Mushroom

    Easy source.

    Just do a search for "and you wont believe" and add all those websites to the list of clickbait creators.

    1. Alan J. Wylie Silver badge

      Re: Easy source.

      Just do a search for "and you wont believe"

      Also, "This one (weird|simple) trick".

      BTW, does an extended regex count as AI?

      1. Mycho Silver badge

        Re: Easy source.

        ...will infuriate you.

      2. Adrian 4 Silver badge

        Re: Easy source.

        'Shocking'

        1. GrumpenKraut Silver badge
          Mushroom

          Re: Easy source.

          More can be gleaned form the subject lines of a certain kind of spam offering random 'survival' items. This apparently targets conspiracy theory nuts. Going like "... that the government wants to make ILLEGAL".

          Real(*) clickbait headlines, I love them just as much as that spam --->

          (*) contrast to playful/funny clickbait on the Reg.

          1. Anonymous Coward
            Anonymous Coward

            Re: Easy source.

            Like this AMAZING LIFE HACK that could save you MILLIONS?

    2. Pete 2 Silver badge

      Free of news content

      It is remarkable how many news "stories" there are, that once you remove all the descriptions, emotional phrases, single-person experiences and advice on what readers should think - once you remove all of that, there is no actual news in the entire article!

      1. Boris the Cockroach Silver badge
        Happy

        Re: Free of news content

        You can do that with politician's speeches too

        Every time the mouth moves up and down you can ignore the sound coming out as it has nothing to do with what the politicians actually does

    3. CHMultimedia

      Re: Easy source.

      Or: Anything with "crazy" "unbelievable" "person quit his job and becomes ticher than the world" "lose 99KG in 2 days" "you won't believe what" "happens next" "%ANY FULLY CAPITALIZED TITLE%"

      Caps are the worst, in my opinion

      1. onefang Silver badge

        Re: Easy source.

        "person quit his job and becomes ticher than the world"

        Person became tighter than the world? Titchier (as in "titchy" meaning very small)? I'd click on that just to find out what they meant.

      2. I ain't Spartacus Gold badge
        Devil

        Re: Easy source.

        That's nothing. I lost 99kg in 1 minute. I shot the mother-in-law.

        I'm here all week!

        [Sorry, couldn't stop myself. 70s flashback. Feeling very ashamed now. Honest.]

        1. onefang Silver badge

          Re: Easy source.

          And your mother-in-law isn't feeling well either.

        2. Jamie Jones Silver badge
          Coat

          Re: Easy source.

          That's nothing. I lost 99kg in 1 minute. I shot the mother-in-law.

          I went on this new "28 day diet".. All I lost was 4 weeks....

    4. Jamie Jones Silver badge
      Windows

      Re: Easy source.

      "10 things you never knew about xxxxxx"

      Argh, that really triggers my OCD. How the hell can they assume they know what I do or don't know on the subject!

      </old git whinge>

  12. Mark M.

    Clickbait headlines vs. intrusive ads

    Never mind the click-baitness of a headline. The AI needs to also gauge the number of unwanted and intrusive ads that hide behind such headlines.

  13. Anonymous Coward
    Anonymous Coward

    If it says "sponsored link"...

    ...then it's probably clickbait.

    1. I ain't Spartacus Gold badge

      Re: If it says "sponsored link"...

      Sponsored links are even worse than clickbait. At least you can hope that a clickbait article might have one single interesting piece of information. Bonus points if it actually turns out to be true...

  14. Bibbit

    Next target...

    Could they point that AI at independent.co.uk or would the AI self-detonate?

  15. Stevie Silver badge

    Bah!

    Oh FFS, the title means nothing, it's the content that is problematical.

    All that will happen now is an AI will be designed to craft anti-AI-spottable headlines for the same tat.

    Said AI will be written by a teenager, in php, over the course of a weekend.

  16. Martin Summers Silver badge

    Any headline image with a red circle around something completely irrelevant. Especially that one about the British woman who got her credit score improved with this 'one neat trick', only she's holding up a FICO report. I've clicked on my fair share of click bait, most of the time for a laugh. It's like giving your brain sweets.

  17. Anonymous Coward
    Anonymous Coward

    pot / kettle

    laughably these articles come with the following recommendations from Outbrain!

    If you are paying more than £5 for your wine you should read this...

    People born between 1948 and 1979 with no life insurance must read this...

    You don't want to read this if you have solar panels...

    How far does £1m go in retirement...

    First ever - breakthrough PPI checker is 100% online - no phone call, no paper...

    1. Martin Summers Silver badge

      Re: pot / kettle

      "First ever - breakthrough PPI checker is 100% online - no phone call, no paper..."

      Got a link?

      1. Jamie Jones Silver badge

        Re: pot / kettle

        Do it properly..... https://www.fca.org.uk/ppi/how-to-check

    2. soulrideruk Bronze badge

      Re: pot / kettle

      Ah, I see you haven't found the joys of adblockers....

      1. Anonymous Coward
        Anonymous Coward

        Re: pot / kettle

        "Ah, I see you haven't found the joys of adblockers"

        Unfortunately this work machine is so screwed down by 'security policies' that I can't install anything useful like that, so I'm constantly amazed by all the crap that gets filtered out at home.

        (I was going to say the settings also screw up sites like Cisco's, but then remembered that I often can't use Cisco's site properly with my 2yo smartphone)

    3. onefang Silver badge

      Re: pot / kettle

      "You don't want to read this if you have solar panels..."

      Oh good, I don't have to read it then.

  18. Anonymous Coward
    Anonymous Coward

    Fancy?

    > Capitalizing the "Loewenstein information gap" is a fancy way of saying exploiting someone's curiosity

    I use that expression all the time, other than in formal settings.

  19. Anonymous Coward
    Anonymous Coward

    Clarification request

    "About eight times out of ten it agreed with humans on whether a title was clickbaity or not, we're told."

    So for the remaining 20%, who was right?

  20. Mark 85 Silver badge

    Academics think differently, obvoiusly...

    "In my opinion, such a thing should not be used in traditional news. But for blogs or articles, which are just meant for light reading and fun –

    He's right they shouldn't be used, but that's ivory tower thinking. Reality... the ad hawkers and yellow sheets will pick it up first (hell, they already do but have to pay humans to do it). Pretty soon, all will be doing it.

    It's pretty scary when ivory tower types publish something with the disclaimer "this should not be used for XXXX". It's like a showing a moth a candle.

  21. Mike 137

    "Artificial intelligent software"

    as opposed, I presume, to real intelligent software.

  22. Anonymous Coward
    Anonymous Coward

    Local Prophet Executed For Blasphemy In Judaea

    You Won't Believe What Happened Next!

    1. Anonymous Coward
      Anonymous Coward

      Re: Local Prophet Executed For Blasphemy In Judaea

      > You Won't Believe What Happened Next!

      Actually, what I don't believe is what they say happened before.

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