AI to re­main our best weapon in the fight against ex­trem­ism, Twit­ter founder says

The National - News - - NEWS - Jack Dut­ton

Ar­ti­fi­cial in­tel­li­gence and be­havioural data will con­tinue to be the best tools to stop vi­o­lent ex­trem­ism on so­cial me­dia in its tracks be­fore it is too late, the founder of Twit­ter has said.

Re­spond­ing to a ques­tion from The Na­tional at the One Young World Sum­mit on Fri­day, Biz Stone, who co-founded the so­cial net­work plat­form with Jack Dorsey in 2006, said that the rise of ex­trem­ism planned on­line was some­thing “the in­dus­try never saw com­ing” when so­cial me­dia web­sites emerged.

“We weren’t think­ing – oh there might be state spon­sored, con­cen­trated co-op­er­a­tive agent work­ing and am­plify an ar­gu­ment that is al­ready hap­pen­ing, and hy­per-po­larise it to the point where it turns into ex­trem­ism. We just didn’t think that and it wasn’t in our heads, so we had to play catch up,” he said.

Pres­sure has mounted on the front-run­ners of so­cial me­dia to quickly re­move ex­trem­ist and dan­ger­ous con­tent, af­ter ter­ror­ists in at­tacks this year – in­clud­ing the Christchur­ch At­tack in New Zealand and the El Paso Wal­mart at­tack – used so­cial net­works to incite ha­tred and broad­cast their at­tacks.

Twit­ter mean­while, has been crit­i­cised for not de­vot­ing as many re­sources to re­mov­ing, or “de-plat­form­ing”, white su­prem­a­cist ac­counts as it has to ISIS ac­counts.

Mr Stone said Twit­ter pre­vi­ously used to be “more re­ac­tive than proac­tive”, re­ly­ing more on hu­man per­son­nel to track down ex­trem­ist con­tent and hate crime, but the plat­form is now “catch­ing up”.

“Now what we can do is re­alise that’s not the fu­ture – we just can’t look at a bil­lion tweets a day that are in Farsi, Ja­panese and ev­ery­thing else. Where we are mov­ing to, is look­ing at be­hav­iour in­stead of con­tent be­cause most of the time cer­tain be­hav­iour is linked to cer­tain kinds of con­tent.”

He said that since Twit­ter adopted AI to crack down on ex­trem­ist con­tent, it has had a pro­found ef­fect and the com­pany now cap­tures around 40 per cent of all dan­ger­ous con­tent im­me­di­ately.

Be­cause of AI and al­go­rithms, Twit­ter can now im­me­di­ately sus­pend ac­counts that ha­rass users, whether a bot or a real per­son, he said.

“I can only imag­ine that it’ll get more and more so­phis­ti­cated as we get more and more data of peo­ple’s be­hav­iours. But not just in­di­vid­ual be­hav­iour but also group be­hav­iour – we need start be­ing more proac­tive.”

He said that most of Twit­ter’s ac­qui­si­tions have been in AI re­cently to help the so­cial net­work­ing site bet­ter iden­tify ex­trem­ist con­tent and re­move it im­me­di­ately.

Biz Stone, left, and Lars But­tler at the One Young World sum­mit

Newspapers in English

Newspapers from UAE

© PressReader. All rights reserved.