WhatsApp Group Join Now
Telegram Group Join Now
Instagram Group Join Now

How Facebook can be more responsible when it goes local.

Tabarez Ahmed Neyazi writes: With ‘internet vernacularization, the platform must invest in understanding local languages, culture, the context in India.

Written by Taberez Ahmed Neyazi |

Updated: November 13, 2021 6:50:23 am

How Facebook can be more responsible when it goes local.
Between 2018 and 2020, several people within Facebook highlighted how hate speech and Islamophobic content have been flourishing in vernacular languages and the inability of artificial intelligence (AI)-based systems, deployed for content moderation, to detect problematic content. (Representational)

Facebook’s internal memos, leaked by whistleblower Frances Haugen — a former Facebook employee — brought out several controversial issues such as rampant misinformation and hate speech in addition to the company prioritising profits instead of investing in understanding local markets. However, what needs the most serious deliberation is the impunity with which Facebook is allowed to treat developing country markets differently than their developed counterparts. This has been illustrated by the relatively minuscule investment by Facebook on screening for fake, inauthentic, polarising and hate-filled content in Indian languages on its platforms. This, when there has been a mushrooming of online content as well as internet users in the Indian languages since the early 2010s.

Between 2018 and 2020, several people within Facebook highlighted how hate speech and Islamophobic content have been flourishing in vernacular languages and the inability of artificial intelligence (AI)-based systems, deployed for content moderation, to detect problematic content. These concerns were brushed aside by the company. While this issue comes across as a classic case of prioritising profits over values, it has a political dimension to it as well. Most of the digital platforms have been operating in India without being required by the regulators to invest in the safety of their users from illegal, obscene and harmful materials.

What was more baffling is the admission by Facebook that it has hate speech classifiers in only four — Hindi, Bengali, Urdu and Tamil — out of India’s 22 scheduled languages. While the company has human moderators to deal with languages other than these four, it only checks content flagged by users. Facebook started focusing on the Indian languages market in 2012 when it launched in eight Indian languages including Hindi and Bengali. By 2017, it was available in 12 Indian languages. However, more than the availability in different languages, it’s important to understand that users are free to share the content in various Indian languages even when they are accessing its platforms in English. Political parties often use vernacular content to reach out to supporters. It’s in this context that Facebook’s approach — to not invest in content moderation in Indian languages — should be taken seriously and further scrutinised.

While Facebook’s revenues from the Indian market have been soaring over the past few years, they are still much lower than the combined revenues from European markets. Social media companies are often driven by profit considerations when they enter a market. They should simultaneously be guided by social values and invest in understanding the societies in which they operate. In fact, they should not be allowed to operate in a market unless they have a comprehensive system in place to factor in culture and context, which, in developing-country markets, are very different from Western and European markets. English language content, which dominated the internet for a long time in India, kept a large number of non-English speaking Indians from accessing the web. With increasing numbers of people accessing social media in their own languages — what I term “internet vernacularisation” — Facebook and other social media companies must invest in building an infrastructure to ensure users are not exposed to harmful and fake content.

Internet vernacularisation revolves around the twin aspects of linguistic diversification as well as the massification of the medium. These two trends have transformed the daily lives of ordinary Indians and have produced unexpected outcomes. This is because the transformations that are occurring in India’s digital space echo what has been occurring in many other countries, in terms of the rise of digital consumption and interactions, digital nationalism and populism and the emergence of multi-lingual, non-English, and vernacular internet spheres. The trend of internet vernacularisation in India is in line with the global trend of the decentralisation and de-Americanisation of the world wide web along with the rise of the global south in terms of the number of web users. The most important contribution of internet vernacularisation is the vast amount of user-generated online content available in various Indian languages. This not only offers new opportunities to understand emerging digital cultures but also provides a window to understand the regional public sphere.

Yet, we cannot ignore the potential downside of the process of internet vernacularisation and the use of digital networks by political elites to advance their self-interest and agenda. Various actors have used the platform to manipulate public opinion and create political polarisation by unleashing targeted propaganda. These attempts have already had dangerous social consequences, contributing to conflicts and killings. While the use of online propaganda has increased with more diverse publics being brought into online platforms, such propaganda does not go uncontested. There are many civil society groups, fact-checkers and advocacy groups — both local and global — simultaneously exposing and busting such propaganda. These contestations among various actors to yield political power and control public opinion are not going to be resolved with internet vernacularisation — they are going to get more pronounced.

It’s in this context, that Facebook — now Meta — which has close to 450 million users in India, needs to be held accountable for the dominant role it has come to play in acting as an important gatekeeper to information. The political class has to act impartially in the larger interests of the society and design regulations that help in checking the abuse of market power, misuse of users’ data and privacy breaches from social media companies. Facebook should not be allowed to operate in a language for which they do not have content moderators. While AI-based algorithms are faster in detecting harmful content, they are not good at understanding Indian languages. Facebook should hire and deploy human moderators directly, rather than outsourcing such jobs, to ensure more accountability from them. In a country that has a long history of communal violence and ethnic tensions, Facebook needs to understand it can be misused to further amplify harmful content from partisan groups and vested interests. Understanding the dynamics of the local informational environment could help in developing grounded strategies and policies that protect the public interest and safeguard users.

 

...