Alphabet’s AI takes up troll hunting
Select language to see a machine translation of this article. The original language of the Article is English and the translation is provided for your convenience.

Alphabet’s AI takes up troll hunting

Recent elections in the U.S. and the UK have made one thing about the internet obvious to even the least online-savvy people in the world: trolls abound. Particularly true in the last American general election, where “alt-right” (alternative right – a term to describe right-wing nationalists whose views are too extreme for the mainstream) commentators lowered the bar for acceptable discourse. At their worst, these people spewed racist screeds and fabricated falsehoods in their support for President Trump.

The visibility of these “trolls” made something else patently obvious: artificial intelligence (AI) alone can’t tell who is a troll and who is not. Or perhaps we should say: it couldn’t tell then.

Alphabet’s share price has risen by 20% so far this year

alphabet 1309

SOURCE: Yahoo Finance

The most obvious example of trolling gone unnoticed came through the propagation of “fake news” from sites like Breitbart.com – a mouthpiece of the alt-right. Here, authors peddled speculation in place of news, and the site’s commentators, many of whom are trolls, made indefensible comments about African Americans and Muslims. The shocking thing, however, was that these pieces of “journalism” were available through respected online news aggregators like Google News and Facebook.

The way these comments slip through “anti-toxicity” filters is simple: AI just isn’t very good at understanding human nuance. Online publication Fast Company uses the following example (sadly, an actual Breitbart comment): “If blacks left America, it would make America great again.” Now, this is clearly offensive to humans – but which words, specifically, are offensive? The answer, of course, is none. To understand why this sentence is toxic (Alphabet’s definition: “a rude, disrespectful, or unreasonable comment that is likely to make you leave a discussion”) requires an ability to interpret the meaning of those words in the context of one another. And AI just can’t do that yet.

But there may be hope on the horizon for a better internet. Alphabet’s Jigsaw division is hard at work on numerous tools that aim to filter out toxic speech online. It’s most famous, and widely used, is called Perspective – although, like all the company’s software, Jigsaw stresses that it’s a work-in-progress.

Currently, the company is building out its software capabilities by exposing it to a huge volume of online hate-speech. The effects of this endeavour remain to be seen, but there is one thing we know for sure: with trolls plaguing nearly every corner of the internet, and human employees lacking the time to address even the smallest fraction of a percent of their vitriol, if Jigsaw can get a functional troll hunter to market, it will be massive.

Disclosure

Dominion holds Alphabet in its Global Trends Ecommerce Fund.


If you would you like to receive the Newsfeeds daily, please click here to sign up now!

Help us make this Newsfeed better by rating this article. 1 star = Poor and 5 stars = Excellent
0.0/5 rating (0 votes)

Disclaimer
The views expressed in this article are those of the author at the date of publication and not necessarily those of Dominion Fund Management Limited. The content of this article is not intended as investment advice and will not be updated after publication. Images, video, quotations from literature and any such material which may be subject to copyright is reproduced in whole or in part in this article on the basis of Fair use as applied to news reporting and journalistic comment on events.