Tech, both high and low, can counter misinformation on Spanish-language media | Opinion

Almost every family has “the one” — the individual who parrots talking points from some conspiratorial Facebook post. Surely they can’t be serious, you think to yourself. Until you find yourself as “the one.”

That’s right, a Ph.D. who has made a career running companies and building technology to prevent this very thing from happening fell for a fake news article. I was enraged and several paragraphs into a heated social-media response before a staff member pointed out I was reacting to Fake News.

If you, like me, grew up in a Spanish-speaking family (mine a Mexican immigrant family in Albuquerque, New Mexico), the chances of having “the one” only increases, given social-media platforms are awash in Spanish disinformation, and countermeasures are woefully inadequate.

This isn’t a niche problem. More than 40 million Americans report speaking Spanish at home regularly. We get news via social-media platforms and use messaging platforms to keep in touch at higher rates than other demographic groups. Organizations. including Equis Institute and the Latino Anti-Disinformation Lab (a joint effort of Media Matters and Voto Latino), regularly catalog the pervasive nature of falsehoods reaching U.S. Latinos through Spanish-language media.

There is an undeniable need for effective countermeasures to ensure our public square, in whatever language, is insulated from efforts to mislead and polarize. And technology, both high and low tech, is critical because a brute-force approach — hiring natural language speakers to moderate content in near real-time — is prohibitively expensive for all but the richest companies.

To see how machines can help, consider how social, dating and gaming companies use artificial intelligence — AI — to police content. “League of Legends” maker Riot Games monitors billions of interactions among players each week, actively enforcing policies that prescribe acceptable behavioral norms. Dating sites owned and operated by The Meet Group use similar tools to monitor 15 million text chats and 50 million direct messages a day for a wide array of toxic behaviors, from misogyny to racism to sexual harassment.

While we can’t simply cut and paste these techniques onto other domains or to simply translate them into other languages, their success at such an extreme scale shows how AI and machine learning can help.

But effective countermeasures can’t stop there — nor does the role of AI. A 2020 study of how machine learning can help combat online disinformation — specifically false information spread intentionally — shows there are detectable patterns within a significant portion of such disinformation. This means platforms can direct resources toward identifying early disinformation trends, as we saw with COVID-19 and election-based disinformation, and deploy AI-based tools to identify patterns in content and behaviors.

While fact-checking false information is crucial, refuting false information fully, rather than just making statements about it, has been proven to be effective. However, a debunk needs to be as memorable as the original false information. Employing not only Spanish-language personnel, but also experts who can provide crucial cultural context, can dramatically boost the success of a debunk. Finally, even simple warnings about the circulation of false information can prepare people to respond better to false information when they encounter it.

To make sure these solutions and others are being adequately resourced and deployed, Big Tech should be required to adhere to stress tests that measure how well they detect misinformation, in Spanish as well as English, perhaps from the Federal Trade Commission or a new digital oversight agency as former FCC chairman Tom Wheeler has suggested.

But not all of the solutions rest in Big Tech’s domain. There is a critical role for traditional Spanish-language media as undergirding all of this is trust and its spillover effects. A 2021 study showed that U.S. Latinos who consumed Spanish-language media reported high levels of trust for Spanish-language journalists, but those who reported consuming mainstream media did not report a corresponding high level of trust for English-language journalists. Most important, trust in Spanish-language journalists positively colored impressions of COVID-related information from public officials.

A final, low-tech countermeasure is to provide more reliable information via Spanish-language radio. Large swaths of Hispanic communities are avid Spanish-language radio listeners. The importance of radio in ensuring a reliable Spanish-language information environment is one of the reasons I recently invested, as part of the largest fundraising effort by a Latina-founded startup, in the Latino Media Network, which is in the process of buying radio stations that reach one-third of Spanish-speakers in the United States.

It is more important than ever to harness the power of technology to ensure tens of millions of Spanish-speaking fellow Americans are getting reliable information online that they take with them into classrooms, doctor’s offices and voting booths. Latinos, including the Spanish-language dominant among us, continue to drive U.S. economic and population growth. We should take every measure necessary to protect us all from the mis- and disinformation that imperil our democratic and open society.

Tom Chavez is co-founder and general partner of the startup studio super{set} and CEO and co-founder of the software company Ketch. He is also the founder of the Ethical Tech Project.