COVID-19: Social media and the spread of information—and misinformation
The coronavirus has led to a deluge of information online. Chatbots and AI on social media platforms are helping to diminish the spread of misinformation.
By Adi Gaskell
June 19, 2020
The outbreak of the coronavirus has led to a deluge of information flooding across the internet. Coupled with the forced time off work for many of us, media consumption has skyrocketed, but this has often been a double-edged sword.
For instance, research from the University of California, Irvine, highlights the damaging impact this media consumption is having on our mental health.
“It’s a public health paradox that has been identified during and in the aftermath of other collective stressors, such as the 2014 Ebola outbreak,” the researchers explain.
In order to avoid this stress, the researchers urge us to avoid speculative stories and to limit our exposure to stories that don’t really provide us with any new information. In other words, quality matters far more than quantity during a pandemic. That is easier said than done, however, as research from the University of Pennsylvania highlights the huge amount of misinformation that’s circling the internet during the pandemic.
The study found that the use of social media to find and consume news content was strongly linked with the consumption of misinformation, whereas people who consumed news from traditional outlets like news channels received more correct information. For instance, people who got much of their news from Facebook and Twitter were far more likely to believe that the coronavirus was deliberately concocted and released by a Chinese laboratory or that the Centers for Disease Control and Prevention had exaggerated the danger posed by the coronavirus in order to damage the Trump presidency.
“Because both information and misinformation can affect behavior, we all ought be doing our part not only to increase essential knowledge about SARS-CoV-2 but also to interdict the spread of deceptions about its origins, prevention, and effects,” the researchers say. “Additionally, all forms of media should ask, ‘Are our audiences better prepared to deal with this coronavirus as a result of our work, or is their trust in us endangering them and their communities?’”
Stopping the spread
Facebook has been using artificial intelligence to try and improve the quality of information shared on the platform since the pandemic broke. The latest Facebook Community Standards Enforcement Report highlights the relative success of its efforts. And, while it reports that 88 percent of the hate speech removed on the site was done so by artificial intelligence (AI) in the last quarter, it’s less clear how successful it’s been in detecting—and removing—misinformation.
The AI-driven approach used on the Facebook and Instagram platforms is driven by natural-language models that aim to fully understand the meaning of any particular post. The models have been trained on huge amounts of data without any human supervision, which the company believes has enabled it to scale up and speed up its offering.
It has also developed AI systems that can analyze content, such as memes, that consist of both images and text. There has been less progress in this regard, not least due to the complexity of the task, but Facebook has recently launched a competition to help it get better at understanding memes.
Its COVID efforts have been a much more human affair and rely primarily on human moderators. The company partners with around 60 fact-checking organizations that attempt to flag misleading information, with the AI then stepping in to try and find similar items to tag in the same way. To date, the company has not been able to train an AI-based system to hunt for misinformation on its own.
WhatsApp has taken a slightly different approach via the widespread deployment of chatbots. Dozens of bots have been deployed on the platform to help ensure that accurate information is relayed to users, but perhaps the foremost among them is the bot developed by the Poynter Institute’s International Fact-Checking Network (IFCN).
The bot is connected to independent fact-checkers in over 70 countries and has a large database of confirmed fake news on the coronavirus. The aim is to allow WhatsApp users to easily check whether any content they’re exposed to on the app has already been identified as false or not.
“Billions of users rely on WhatsApp to stay in touch with their friends and families every month,” IFCN’s Baybars Örsek says. “Since bad actors use every single platform to disseminate falsehoods, to mislead others during such troubling times, fact-checkers’ work is more important than ever. The IFCN chatbot will allow users to search for fact-checks and get connected with fact-checkers in their countries from their smartphones. The chatbot will also serve as a way to direct people to their local fact-checkers’ websites.”
The bot, which is free to use, is accessed by saving +1 (727) 2912606 as a contact number. Users then simply need to text the word hi to the bot to activate a conversation with it. The bot has been designed to be user friendly and consists of a short numerical menu.
The UK government has also been utilizing chatbot technology via WhatsApp, with over one million messages sent to the chatbot since it was launched at the start of April. As with the IFCN chatbot, the tool aims to provide citizens with official, trustworthy, and timely information about the coronavirus and thus to combat the spread of misinformation.
“People all over the world are using WhatsApp more than ever to connect with and support their friends, family, and communities during this crisis,” WhatsApp’s Matt Idema says. “Now, thanks to the UK Government’s Coronavirus Information Service, the British public can also get official, timely, and trustworthy advice about coronavirus on WhatsApp. We will continue to support the UK government to develop the service with updated information and official guidance as the situation develops.”
The app has expanded to offer information on a wide range of coronavirus-related topics, with a particular focus on misinformation. These topics include a deep dive into social distancing, advice for those who are pregnant, and information on applying for Universal Credit and other financial support. The service is free to use and is accessed by adding the +44 07860 064422 telephone number to one’s contacts and texting the word hi to begin the conversation.
In the United Kingdom alone, there are 22 coronavirus chatbots available for WhatsApp, including the World Health Organization’s WHO Health Alert, which was launched last month and had received over 10 million users within three days of launch.
The coronavirus pandemic has seen a huge digital transformation across society but especially in healthcare, with telehealth and remote monitoring increasingly common. It will be interesting to see whether chatbot-based health services will grow in unison, as millions of people get used to accessing accurate and reliable health information via such a medium. In the meantime, the army of chatbots on WhatsApp is performing an invaluable service in diminishing the spread of misinformation online. And, for now, that’s a good enough outcome.