Objectivity in search: New Bing feature tells you many sides of the story

By Joanne Kuai

Objectivity in search: New Bing feature tells you many sides of the story

//
Posted By
/
Comment0
/

By Joanne Kuai

London, 2nd, November 2018

 

A new Bing feature, spotlight, is launched to provide overviews of news topics in search results. Now, when searching on the search engine developed by Microsoft, the new AI-driven feature with human intelligence assistance offers readers the latest headlines, a rundown of how the story has developed over time, relevant social media posts and multiple perspectives.

“Transparency and trust in journalism are in crisis. Confirmation bias is also critical when technology is becoming more and more sophisticated and personalized. We want to address these issues.” said Natalie Mao, Director of Product Management of Bing News at Microsoft.

“When you search something online, you begin with your own perspective. Sometimes you just don’t know what you missed. In other words: you don’t know what you don’t know!”

Ms. Mao gave an example of Bing innovation of providing multiple perspectives to an answer – when you search for “are video games good for you,” the results will not only show you articles on “playing video games can be bad for your health” but also “playing video games can be good for your brain”. Ms. Mao also gave another example of searching “electric cars” – Bing news spotlight experience provides an overview of “electric cars” topic right in the Bing search results with headlines, rundown, and multiple perspectives from reputable sources.

“We believe it’s our responsibility to try to give people a more comprehensive, balanced and informative picture when they are searching for information on Bing,” said Ms. Mao on Friday at the 8th Data & AI for Media Conference in London.

The conference is for media and data practitioners who are seeking to earn revenue and expand audiences by leveraging data and artificial intelligence. The 2-day event features a powerful format of general sessions with top-notch speakers, plus interactive, small-group breakout sessions called “huddles.”

 

Trustworthy journalism

Microsoft’s Bing news spotlight aims to help users “learn more about news in less time”. For major developing news or controversial topics, the multi-perspective news feature shows different viewpoints from high-quality sources. It also features “Fact Check” label and trust indicator to identify the different genres, among news report, review, commentary and opinion etc. in its news search results. For a source to be considered high quality, it must meet the Bing News PubHub Guidelines – a set of criteria that favours newsworthiness, originality, authority, and readability.

In a bid to “get people out of their bubbles”, the AI-driven feature assisted with human intelligence initially launched in the US in August is now available across the pond. Microsoft announced the launch of Bing spotlight in the UK in early October.

Though whether Bing could achieve its goal through the new feature is yet to be examined, the well-intentioned initiative has received positive reviews from users and is welcomed by the journalism field. Deputy editor of NiemanLab at Harvard University, Laura Hazard Own, calls it “an interesting example of how a search engine is trying to tackle news aggregation in an explicitly responsible way”.

 

Users at heart

This new feature is also derived from the knowledge of users’ needs and pain points. Through a user study, Microsoft learns that perception of bias is one of the top online news consumption pain points and Ms. Mao says users are now loving the multi-perspective news spotlight feature, quoting a user saying the spotlight gives “a more in-depth news experience”.

Though the latest data from web traffic analysis tool StatCounter puts Bing’s search engine market share worldwide across platforms at 2.17% in October 2018, and at 8.58% on Desktop in the US, a 3.68 percentage point drop from a year ago, Microsoft claims otherwise.

It says the number of users at Bing continues to grow in the recent years, with Bing Network PC share increased to 34% in January 2018 from 31% in December 2015 in the US, based on statistics of American media measurement and analytics company comScore. It also claims a 7-percentage point increase to 25% during the same time period in the UK market

In addition, comScore’s data from March 2018 shows that Bing users in USA are balanced in gender, with around 50% female and 50% male. Half of Bing users are under the age of 45. Half of Bing Network audience graduated from college. And nearly one-third of Bing Network audience has a household income of $100,000.

 

Responsible AI

AI innovation is in all aspects at the tech company. Ms. Mao, who is also a member of the Artificial Intelligence & Research Group at Microsoft, introduces one of Microsoft’s AI advancements: Text to Speech (TTS), a technology that allows written text to be output as speech and where Microsoft is in the lead as the first one to reach human parity.

Microsoft’s Neural TTS is a joint optimization of pronunciation and prosody, plus high-fidelity audio generation, learning from large datasets across speakers. During the presentation, more than half of the participants failed to tell the difference between the recorded human voice from synthesized ones. Besides the highly natural TTS voice, Microsoft also provides the Custom TTS solution which helps 3rd party companies to build their own branded TTS voices.

Ms. Mao gives an example of Microsoft’s partnership with Chinese AI solution provider Roobo. Using Microsoft’s Custom Voice, which creates natural brand TTS voices for robots, Roobo is able to develop an English tutoring app for kids First Leap. “In China, English tutors that are native speakers can be quite expensive. A partnership of this kind makes education more affordable and accessible for the people in need,” says Ms. Mao.

In the Future Computed published by Microsoft this year, six key principles guiding the development of AI are outlined. There are fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability. Ms. Mao says these principles are foundational to how they approach AI at the company.

 

 

 

 

 

 

 

Leave a Reply