Machine Learning Generates Clickbait Headlines That Will SHOCK You (4)

Norwegian developer and blogger Lars Eidnes has designed a clickbait generator using a neural network, which is able to create sensationalist headlines that play on human readers’ curiosity.

The internet is littered with clickbait articles, which inflate insignificant news and events to catch a reader’s attention. A growing number of popular publishers, such as Buzzfeed, Jezebel and Upworthy, have all been accused of the tactic. The technique often features emboldened text, capital letters and promise of items that will shock and amaze (i.e. “16 Facts You’ll Never Believe Are True…”)

Eidnes trained his neural network by scanning around two million clickbait titles from online media sites. When asked to form a sentence, the system can now output a single word and continues the prediction process to find related words, in a pattern known as Recurrent Neural Networks (RNNs).

After the first training, the clickbait generator was still incomplete, producing sentences such as ‘Real Walk Join Their Back For Plane To French Sarah York’ and ‘2 0 Million 9 0 1 3 Say Hours To Stars The Kids For From Internet.’ However, following further training rounds it was able to come up with grammatically correct titles including ‘Kate Middleton Looks Into Marriage Plans At Charity Event’ and ‘WATCH : Gay Teens Made Emotional Letter.’

Eidnes has created Clickotron.com, updated with artificially created news stories every 20 minutes. Each article’s image is found by searching the Wikimedia API with the headline text, and selecting pictures with a permissive license. Like many online platforms, the site uses a voting system to rank the articles.

He concludes his blog post: “…this gives us an infinite source of useless journalism, available at no cost. If I remember correctly from economics class, this should drive the market value of useless journalism down to zero, forcing other producers of useless journalism to produce something else.”

…Ouch!