‘Bots’ step up for 2016 elec­tion news cover­age

Kuwait Times - - TECHNOLOGY -

If you’re read­ing about the US elec­tion, some of that news is likely to come to you from a “bot.” Au­to­mated sys­tems known as “bots” or “robo-jour­nal­ism” have been around for years, but they are play­ing a big­ger role in cover­age this year amid tech­nol­ogy ad­vances and stretched me­dia re­sources.

The New York Times, Wash­ing­ton Post, CNN, NBC, Ya­hoo News and the non-profit Pro Publica are among news or­ga­ni­za­tions us­ing au­to­mated tech­nol­ogy or mes­sag­ing bots for cover­age in the runup to Tues­day’s vote or on elec­tion night. News or­ga­ni­za­tions are in­creas­ing use both of sys­tems which em­ploy al­go­rithms to cre­ate text from data, and of au­to­mated “bots” de­liv­er­ing up­dates to smart­phones.

The New York Times bot on Face­book Mes­sen­ger launched ear­lier this year uses short dis­patches from re­porter Nick Con­fes­sore, and al­lows users to in­ter­act to get de­tailed bot-driven news up­dates or polls. “This is a nat­u­ral fol­low-on to what we have been do­ing in con­ver­sa­tional jour­nal­ism,” said New York Times prod­uct di­rec­tor An­drew Phelps.

Those sign­ing up for the bot re­ceive pe­ri­odic short mes­sages such as, “Hey it’s Nick. The race took a swerve this week­end.” Con­fes­sore said it was chal­leng­ing to find the right for­mula for “a re­ally short form of sto­ry­telling,” in a text mes­sage, but noted that the bot al­lowed read­ers to “drill down fur­ther” to get more de­tails.

Meet­ing peo­ple on plat­forms

Phelps said bot us­age has been “in six fig­ures” with an au­di­ence that is younger and more global than Times read­ers. “It’s an ef­fort to meet peo­ple on mes­sag­ing plat­forms,” he said. “We wanted to make it more per­sonal, more in­ter­ac­tive, to al­low read­ers to feel more con­nected to the jour­nal­ists them­selves.”

While bots of­fer no im­me­di­ate mon­e­ti­za­tion, it can helps bring more peo­ple to the news­pa­per’s apps and web­site. “This gets at the heart of re­la­tion­ship build­ing,” Phelps said. The Wash­ing­ton Post mean­while has its own bot us­ing a ro­bot icon, and separately will be us­ing ar­ti­fi­cial in­tel­li­gence tech­nol­ogy to up­date dis­patches on elec­tion night.

The Post, owned by Ama­zon founder Jeff Be­zos, de­vel­oped the sys­tem called He­li­ograf that will help cre­ate hy­brid hu­man- and com­puter-gen­er­ated sto­ries. He­li­ograf al­lows the Post “to cre­ate sto­ries that are bet­ter than any au­to­mated sys­tem but more con­stantly up­dated than any hu­man-writ­ten story could be,” said Jeremy Gil­bert, di­rec­tor of strate­gic ini­tia­tives at the news­pa­per.

Buz­zFeed ex­per­i­mented with its “Buz­zBot” on Mes­sen­ger dur­ing the 2016 po­lit­i­cal con­ven­tions as part of an ef­fort to bet­ter con­nect with read­ers and par­tic­i­pants at the events.

Amanda Hick­man, who heads the Buz­zFeed Open Lab, said bots of­fer “a one-to-one re­la­tion­ship which gives us the op­por­tu­nity to let peo­ple fine-tune the prod­uct they want.”

Turn­ing data into sto­ries

Al­go­rith­mic sys­tems which turn data into news sto­ries have been used for sev­eral years, mainly for rou­tine cor­po­rate re­sults and mi­nor league sports, but are now play­ing a role in elec­tion cover­age.

The non­profit news site ProPublica’s elec­tion data bot, cre­ated with Google News Lab, up­dates ev­ery 15 min­utes with elec­tion fore­casts, cam­paign fi­nance re­ports, Google Trends and other data. An­other non­profit called the Pol­lyVote Project de­liv­ers sim­i­lar dis­patches based on poll re­sults and other data.

“Ev­ery time we get new data we cre­ate an au­to­mated news item,” said An­dreas Graefe, a fel­low at Columbia Univer­sity’s Tow Cen­ter for Dig­i­tal Jour­nal­ism who is leader of the project funded by the Tow Cen­ter and Ger­many’s LMU Mu­nich Univer­sity. “We can pub­lish ar­ti­cles sec­onds af­ter we re­ceive the data, and we can do it in an un­lim­ited amount.”

Graefe said sev­eral stud­ies in Europe have shown that read­ers can­not tell the dif­fer­ence be­tween a hu­man- or com­puter-gen­er­ated news ar­ti­cle. “When you ask peo­ple how read­able a story is, they rate the hu­man ar­ti­cle bet­ter, but when you ask them how cred­i­ble it is, the com­puter is bet­ter,” he said. “We don’t re­ally know why.” — AP

Newspapers in English

Newspapers from Kuwait

© PressReader. All rights reserved.