International

WHO launches pilot of AI-powered public-access social listening tool

06/02/2021

29 January 2021

What is the pilot and why is it important?

If you’ve been following WHO’s infodemic management updates and training courses at any point over the last year, you may have learned that one of the fundamental principles of responding to infodemics is understanding where they originate and how they spread.

Since the early days of the COVID-19 pandemic, WHO’s infodemic management team has worked to develop analytical approaches including with the use of artificial intelligence (AI) to help identify rising narratives that are catching people’s attention in online conversations. This effort helps health authorities better understand what information people are seeking so they can meet that need.

At the same time, it has been important to try to identify so-called “information voids” where there’s a lack of credible information to feed into conversations and answer people’s questions. When people face delays getting the information they need, it can lead to a rise in speculation and conspiracy theories, and that too is an infodemic challenge that can result in harm to people’s health.

One strategy that health authorities can deploy to counter the infodemic is to proactively identify both rising topics and information voids so they can saturate online conversations with high-quality health information that responds to the questions and concerns of the public.

But how do you listen to conversations on the internet to find out what’s concerning the people you serve—especially when governments and public health cannot afford investment in infrastructure and software?

WHO has built a platform called EARS, or the Early AI-supported Response with Social listening tool, which allows health decision-makers to view a real-time analysis of narratives happening in public online forums in multiple countries and languages.

So how does EARS work?

Let’s say you work for the Ministry of Health and you start noticing some posts about non-proven cures for COVID-19, and you want to know if this is a one-off instance or it’s a topic that’s gaining traction more widely. Just go to the EARS dashboard where you can pull up a list of topics that are trending—or are on the rise—in any of the pilot countries and around the world.

The AI platform that powers EARS is combing all kinds of publicly available online sources. It’s pulling publicly available data from the major platforms—Facebook, Twitter, Instagram, etc.—but it’s also mining comments on publicly available blogs, news articles and online forums to see what people are talking about based on keywords and hashtags.

And because the platform interprets natural language processing (NLP), it picks up on context cues about people’s intent behind their posts, like whether they’re asking a question or they’re complaining.