Trends in News­rooms 1: The Rise of Bots

Newspapers & Technology Magazine - - Industry Insight - ▶ Spe­cial to News & Tech FROM Si­mone Flueck­iger si­mone .flueck­iger @wan -ifra .org

Chat­bots are on the rise within the news in­dus­try, with many or­ga­ni­za­tions ex­per­i­ment­ing with the Ar­ti­fi­cial In­tel­li­gence tech­nol­ogy. But what's be­hind this trend, and what ben­e­fits can chat­bots re­ally bring to a me­dia out­let? It’s been more than a year since Quartz launched its nascent chat app that sparked in­ter­est, praise and crit­i­cism within the in­dus­try. And ever since com­pa­nies like Face­book, Tele­gram or Kik made the tech­nol­ogy to build chat­bots avail­able to the gen­eral pub­lic, news or­ga­ni­za­tions have been ex­per­i­ment­ing with it in var­i­ous forms. Most news or­ga­ni­za­tions see this trend as just that: a chance to ex­per­i­ment with an emerg­ing tech­nol­ogy on es­tab­lished plat­forms. And clearly, these ex­per­i­ments are pri­mar­ily cen­tered on test­ing the wa­ters of a more per­son­al­ized ex­pe­ri­ence with their users. But like with any new tech­nol­ogy, chat­bots come with the ex­pected yin and yang.

A more per­son­al­ized news ex­pe­ri­ence

“I think one of the big ad­van­tages is the abil­ity to per­son­al­ize your news stream to your taste and to your needs,” said An­drew Haeg, CEO of GroundSource, a com­pany that al­lows or­ga­ni­za­tions to build chat­bots.

“For­ever, we’ve created sto­ries that peo­ple re­ceive and they’re largely un­dif­fer­en­ti­ated and ev­ery­one re­ceives the same story. But if you have a dif­fer­ent level of ex­per­tise or cu­rios­ity about the story, you don’t re­ally have a way of cus­tomiz­ing the story to your own needs.”

GroundSource re­cently teamed up with the Texas Tri­bune to help them de­velop its Face­book Mes­sen­ger bot, Paige, aimed at de­liv­er­ing in­for­ma­tion about the Texas leg­is­la­ture.

With Paige, which sends out up­dates on Mon­days and Fri­days and al­lows users to ask ques­tions or sub­mit tips, the Tri­bune is try­ing to reach an au­di­ence that is cu­ri­ous about pol­i­tics but not too heav­ily in­vested in the sub­ject mat­ter.

“They had a re­ally good, vivid pic­ture of who they were serv­ing with Paige,” Haeg said.

“When you’re set­ting out to cre­ate a bot, you re­ally need to know who it’s for and what in­for­ma­tional needs those peo­ple have. Do­ing ac­tual re­search about who the bot is for will in­form ev­ery­thing from the voice of it to the in­for­ma­tion it’s go­ing to con­vey, to the ques­tions you can ask it, as well as the mar­ket­ing and the dis­cov­ery of the bot.”

With 1,000 peo­ple sub­scrib­ing to the bot within two weeks, Haeg noted, “early in­di­ca­tions are re­ally strong that peo­ple want to en­gage in this way.”

Boost­ing dis­cov­ery of con­tent

Rap­pler, a dig­i­tal pure player in the Philip­pines, launched its Face­book Mes­sen­ger bot in a bid to stay con­nected with its user base that was mov­ing from Twit­ter to Face­book, and to im­prove dis­cov­ery of its con­tent.

The chat­bot, dubbed RapRap, func­tions as a search and con­ver­sa­tional tool, sur­fac­ing con­tent re­lated to key­words, and al­low­ing users to ask ba­sic ques­tions.

“They wanted to in­crease dis­cov­ery of their con­tent,” Kevin An­der­son, who cov­ered Rap­pler as part of his re­port “Be­yond the Ar­ti­cle”, ex­plained in a WAN-IFRA we­bi­nar.

“They found that sim­ply shar­ing their con­tent on Face­book was re­ally lead­ing to kind of a su­per­fi­cial level of dis­cov­ery, and there are cer­tain things that work well on so­cial plat­forms, like very emo­tive con­tent, and iden­tity-based con­tent. But they re­ally wanted peo­ple to be able to see and for the chat­bot to sur­face the widest range of con­tent to their read­ers.”

The chat­bot has also al­lowed Rap­pler to bet­ter com­mu­ni­cate its ed­i­to­rial pri­or­i­ties with its au­di­ence, since the Face­book news­feed of­fers few pos­si­bil­i­ties to high­light or dif­fer­en­ti­ate be­tween the var­i­ous types of con­tent Rap­pler’s staff pro­duces.

“Peo­ple see an ar­ti­cle but there’s no sense of the im­por­tance of the ar­ti­cle, it doesn’t give you the same kind of ed­i­to­rial pri­or­ity you have with the de­sign of your own web­site or push no­ti­fi­ca­tions,” An­der­son said.

“They also wanted to com­mu­ni­cate the types of con­tent they pro­duce, news, com­men­taries or anal­y­sis. And they thought that was also lost in the news­feed. They felt that with a chat­bot they could com­mu­ni­cate the dif­fer­ent types of con­tent and ways they pro­duce them.”

Ad­di­tion­ally, Rap­pler built a chat­bot to help with crowd­sourc­ing in­for­ma­tion for their projects around cor­rup­tion re­port­ing or good gov­er­nance, al­low­ing users to sub­mit re­ports.

A shift from open so­cial net­works to closed mes­sag­ing apps

While the launch of RapRap was driven by a change in so­cial me­dia tastes, users mov­ing from open so­cial net­works to closed mes­sag­ing apps has also played a part in the rise of chat­bots. In fact, a Busi­ness In­sider In­tel­li­gence re­port found that the com­bined user base of the four big­gest chat apps (What­sApp, WeChat, Vibr and Face­book Mes­sen­ger) has al­ready sur­passed that of the four big­gest so­cial net­works (Face­book, Twit­ter, In­sta­gram and Snapchat).

Mes­sag­ing ser­vice Tele­gram rolled out its bot API and plat­form in 2015, and was used for the launch of Politi­bot, a chat­bot created for the Span­ish elec­tion last year.

“We chose Tele­gram be­cause it was heav­ily used by politi­cians in Spain, es­pe­cially from the new par­ties,” ex­plained co-founder Ed­uardo Suarez.

“We thought it was a good way to con­nect with po­lit­i­cal junkies.”

Launched a day be­fore the elec­tion cam­paign started, the bot served up daily news di­gests, which in­cluded graph­ics, long and short-form ar­ti­cles, and pro­vided users with real-time elec­tion up­dates within their con­stituency and on a na­tional level.

Within three weeks of the bot go­ing live, Suarez said 8,000 users had signed up, and 103,510 ses­sions were recorded out of which one-third lasted longer than five min­utes, and two-thirds longer than one minute. On top of that, weekly user re­ten­tion stood at 56 per­cent.

Nat­u­rally, the bot’s ca­pa­bil­i­ties were limited and it couldn’t un­der­stand all ques­tions and mes­sages it re­ceived, but Suarez said that ac­tu­ally helped spread the word about it.

“Peo­ple were con­nect­ing with the bot. Some of its an­swers to ques­tions it didn’t un­der­stand went vi­ral,” Suarez said.

“From the be­gin­ning, we said this is a very clumsy robot, this is an ex­per­i­ment, and we tried to lower peo­ple’s ex­pec­ta­tion in that re­gard, and I think that is a good idea.”

The bot is now also avail­able on Face­book Mes­sen­ger, cov­er­ing in­ter­na­tional pol­i­tics in ad­di­tion to the Span­ish po­lit­i­cal land­scape.

Lim­i­ta­tions of chat­bots

The above ex­am­ples show in what ways chat­bots can be used to ben­e­fit a me­dia or­ga­ni­za­tion, from reach­ing new au­di­ences to pro­vid­ing more per­son­al­ized ex­pe­ri­ences. But the tech­nol­ogy is in many ways still limited.

“One of the most in­ter­est­ing things is that peo­ple re­ally do chat to them as if they were liv­ing sen­tient be­ings and the tech­nol­ogy is nowhere

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.