Twitter's Automated System Is Making Mistakes In Labeling Tweets Spreading COVID-19 Misinformation

Twitter began labeling tweets spreading a conspiracy theory that 5G is the cause of the coronavirus on May 11, and its automated technology is making mistakes while reviewing content. According to authorities, some individuals started to set fires to cell towers who think that the false theory is true. Twitter will take down posts that spread coronavirus misinformation. However, if a tweet does not contain the same level of harmful information, it will be labeled with a link that redirects people to a trusted source of information.

The automated technology has made a lot of mistakes including labeling tweets that refute the fake theory and contain the correct information, and the technology has also labeled tweets sharing links to news stories from trusted sources such as BBC, Reuters, Voice of America, and Wired related to the conspiracy theory. According to experts, mislabeling may confuse people, particularly when a user does not tap the label. According to a computer science professor at the University of California, Berkeley, Hany Farid, mislabeling the tweets is doing more harm than not labeling the tweets at all.

The platform declined to show how many tweets containing 5G-coronavirus information Twitter has labeled and did not provide any estimated error rate. Twitter says that the Trust and Safety team is keeping track of the labeled tweets. Moreover, Twitter has not yet fixed those mislabeled tweets identified by CNET. Twitter stated that the company is improving its automated systems.

Twitter plans to tackle other hoaxes as well and the platform is facing massive moderation challenges with 166 million daily active users who are monetizable. Twitter states that its automated systems are helping the employees to efficiently review reports. The company’s approach to COVID-19 misinformation is similar to that of Facebook’s efforts to tackle misleading information. Although Facebook relies more on Human resources to review content and operates with over 60 third-party fact reviewers across the globe, Twitter is using automated systems to label content without a human reviewing the content first.

Farid was not surprised to see Twitter’s systems making errors and stated that the platform could take actions against users spreading misleading information. According to a study released by researchers at Oxford University during April, high-profile users shared around 20% of false claims and generated around 69% of the total social engagement. Some users are testing Twitter’s systems by using the words coronavirus and 5G in their tweets.

Ian Alexander, a YouTube content creator spotted that the systems automatically label tweets if a person types 5G, coronavirus, or COVID-19 in a tweet. MIT’s study discovered that users may believe unlabeled information even if those stories contained misleading information. A professor at MIT Sloan School of Management, David Rand states that one solution for social media companies is to ask their users to rate the content as accurate or inaccurate.


Photo: SOPA Images via Getty Images

Read next: Elon Musk realizes his mistake - Wants to take back some of the tweets made during COVID-19 lockdown

No comments:

Post a Comment