Conversational bot will help counter false news during Brazil’s elections

Brazilian voters will have a robot helper to fight disinformation in this year's general elections. Her name is Fátima, a conversational bot that is being developed by the team at fact-checking site Aos Fatos in partnership with Facebook. Fátima’s launch is scheduled for June, in time for the October elections.​

Through Messenger, Facebook's instant messaging service, the bot will provide conversations with tips on how to analyze information published online. Fátima– whose name is a play on "FactMa," an abbreviation for Fact Machine– will suggest, for example, that readers check if a news item has been published by a known journalistic website or if the language used in the text is adequate for journalistic standards. News consumers will learn how to separate news from opinion, how to find reliable data for various topics and how to determine if a source is reliable, according to a release from Aos Fatos.​

"Fátima will never say that information is false. The goal is for her to instruct people to draw their own conclusions," Tai Nalon, executive director and co-founder of Aos Fatos, told the Knight Center. 

Initially, the Aos Fatos team crafted a browser plugin that linked keywords to fact-checking links made by the site. However, the project did not move forward due to a lack of funding, and they decided to test a more ambitious path, according to Nalon. Conversational bots from news organizations, although more widespread in the United States, are still rare in Latin America and especially in Brazil.​

The idea is for Fátima to talk to users– anyone who goes to the Aos Fatos page can contact the bot. Currently, the project’s editorial team, made up of Nalon and journalists Sérgio Spagnuolo, Bárbara Libório and Ana Rita Cunha, is working to develop the bot's responses based on manuals on best fact-checking practices. The methodology is verified by the International Fact-Checking Network. The team also conducts studies to understand how Brazilians consume news and what readers' main doubts are, in order to adjust the chatbot’s responses.

On the other hand, the technology team has the challenge of adapting the technology available in English on conversational bots into Portuguese. The technology director and co-founder of Aos Fatos Rômulo Collopy added computer scientist Marco Rougeth, a specialist in bot development, and developer Ana Schwendler, a specialist in natural language processing, to his team.

According to Nalon, the next step after the release in June will be to allow bot responses to offer links to the latest checks on any particular topic. The journalist still dreams of making a similar tool for WhatsApp, an application with 120 million users in Brazil

"There are technological impediments to WhatsApp. The API (application programming interface) is not open. It is still very difficult to have a dialogue with WhatsApp, but it would be very important to combat disinformation," she explained.

The project is part of a news literacy initiative funded by Facebook in Brazil. Another project supported by the company is the online course Vaza, Falsiane! (Get out, phony!) The purpose of the classes is to help teens, youth and educators identify and prevent the spread of false news.

On a Facebook page, the course will publish short videos with the participation of public figures, as well as quizzes, lists and memes. The creation is from Ivan Paganotti (from the University of São Paulo - USP), Leonardo Sakamoto (Pontifical Catholic University of São Paulo - PUC-SP) and Rodrigo Ratier (Casper Líbero College). Sakamoto is also director of the NGO Repórter Brasil and Ratier is executive editor of the site Nova Escola.

"Being able to identify false news is a fundamental skill in today's world, as well as to recognize the different degrees of misinformation from false news to fabricated truths and skewed data," professor Paganotti said, according to a Facebook press release.​

Both projects were suggested at a roundtable organized by Facebook in September of last year with academics, media education experts and journalism associations, with the goal of making projects to form a more informed community. ​

Social media should play a key role in this year's elections, where Brazil's presidents, governors, deputies and senators will be chosen. Due to the electoral law passed in 2017, social networks and search tools are the only online environments where parties and candidates can pay to push content.

According to figures from USP's Public Policy Research Group for Access to Information (Gpopai), about 12 million people shared fake news about online politics in Brazil in June of last year. In October of last year, the Higher Electoral Court announced the development of a task force to fight lies on the internet.

For Nalon, this scenario poses a challenge for professional journalism on social networks. Therefore, it is important to use technology as a tool in favor of quality information and against producers of false content. "There is no way to fight a war against someone who has weapons of the latest generation while we are using bow and arrow," she said.

Last year, Facebook and other major technology companies were taken to task for their roles as platforms for spreading false news during the U.S. presidential elections. At hearings at the US Congress in November, executives from Google, Facebook and Twitter admitted that Russian operators used digital platforms as tools to divide the country. Facebook is still testing out methods to stamp out false stories disseminated on the platform. On Dec. 20, it announced it would no longer use Disputed Flags to tag these pieces, but would instead offer Related Articles “to help give people more context about the story.”

"Faced with a flood of accusations against digital platforms, they really needed to encourage initiatives like this," she added. "There is no silver bullet, a vaccine to end this pandemic of misinformation. But if misinformation is concentrated on these platforms, the responsibility [they have] is proportional."

Note from the editor: This story was originally published by the Knight Center’s blog Journalism in the Americas, the predecessor of LatAm Journalism Review.