The War of Fake News and How AI Can Help Combat It

Internet’s ubiquitous influence has raised both suspicious eyebrows and loud voices. Beyond simple combinations of words and sentences lie serious acts of internet abuse, division, communalism and propaganda of all kinds. One such menace, the Dissemination of Fake News is gaining momentum and nearing the form of an epidemic. With over 3.2 billion internet users across the world, 699 million users spring from India alone. While the “culture” of Fake News is an alarming threat to the sanctity of internet users, it increases exponentially during elections, campaigns and during incidents of mass agitation and disruption in law & order.

In a report published by Oxford University, the researchers have expressed serious concern over the misappropriation of social media platforms and private messaging applications, particularly WhatsApp, to carry out the malicious business of misinformation. Social Media manipulation is a big business. The internet becomes a hub of fake videos, images, incidents, and content that stresses on things said and done that originally never happened. This content is psychologically targeted, extremely misleading, and provokes the sentiments of the masses. Because the objective of fake news is to ‘spread like wildfire’, the contents of such posts increase magnanimously and intend to support/oppose an ongoing revolution.

The advancements in technology now enable users to create and curate fantastically fake content, especially videos and pictures that prima-facie look realistic. Legitimacy is seriously gambled and one fails to identify with the naked eye if the videos are forged or real. Thus, the struggle to differentiate and distinguish between the real and the fake is real. The prominent rise in fake news is often termed as rise in “Deep-Fakes”, with manipulative content.

In the Indian context, the latest BBC article deems WhatsApp, the private messaging platform as a “blackhole” of fake news in India’s election. India has a whopping 82% of users on WhatsApp. With the diverse ethnicity and religions practiced in the country, the consequences of fake news become proportionally disastrous. The internet usage in semi-urban and rural areas has increased exponentially, leading to increased circulation of misleading information, accounted by low-literacy rates and lack of awareness. Furthermore, a majority of 52% of smartphone users consider WhatsApp and Facebook as the holy grail of news. This is a worrisome finding that highlights the core of the problem of fake news, where users take forwarded content prima-facie. In addition to this, the 36% of respondents of this survey have low trust in news sites while interestingly express higher levels of trust in news in search (45%) and social media (34%). The statistics from a highly complex market of Indian internet users aggravate the menace of fake news in fueling riots, protests, bandhs, communal killings and rigged elections.

Can AI combat the menace of fake news?

The answer is affirmative. AI has achieved considerable milestones in detecting digital deep-fakes, especially videos and pictures. Various models are trained over huge datasets to ascertain and identify the fabricated content. Similarly, for text-based content, AI makes use of linguistic tools to spot fake news. Linguistic signs like grammar, word-choice and punctuation along with exaggeration and exclamation are often analyzed to spot fake news.

In addition, there is a dire need of initiatives by social media giants to combat fake news. As a sigh of relief,Facebook owned WhatsApp introduced the “Forward” label to identify the forwarded messages, further limiting the number of forwards by a single user to five. WhatsApp, Twitter and Facebook have banned several hundred users, accounts, pages, and groups spreading misinformation during crisis.

Fact-checkers at various news houses and social media houses are in great demand. While machines are trained to detect manually curated fake content, the bigger challenge is to detect a machine-created fake content by AI. The automation of various parameters to detect fake news is a growing trend, such as spotting the source of the published/circulated article, analyzing the network of the informants spreading the article and picking linguistic cues to identify fake text. While AI cannot entirely replace manual investments in detecting fake news yet, it can definitely complement and accelerate this process.