Will AI be the Nemesis of ‘Fake News’?

Deliberate dissemination of inaccurate news to spread panic and confusion has been the bane of excessive information sharing in the last five years. From creating political bias in order to swing election results to causing panic in the world during the COVID-19 pandemic, fake news has an abiding impact on social and political issues.

Top Machine Learning and AI Courses Online

There is a growing need to have a strong nemesis against fake news. Sophisticated AI tools have been stepping up to run fact-checking algorithms to offer authentic news and reliable information to the world. But will they be able to eliminate fake news altogether? This is a question that data scientists are still grappling with.

Introduction

2019 was an eventful year for India with the general elections, an appalling attack against security personnel in Jammu & Kashmir, abrogation of Article 370 and the nationwide protests against the Citizenship Amendment Act dominating national news and media. An equal and extensive distribution of misinformation was doing the rounds, thanks to WhatsApp, Twitter, Facebook and other social media channels.

Trending Machine Learning Skills

Enrol for the Machine Learning Course from the World’s top Universities. Earn Masters, Executive PGP, or Advanced Certificate Programs to fast-track your career.

An article in the Economic Times (Link to: https://tinyurl.com/w86mclp) marked 2019 as ‘the year of fake news’ with fact-checking teams on their toes for almost the entire year. The impact of fake news on society did not amount to merely ideological and political differences. In many cases, it was as adverse as killing minorities and mob violence against them. As the second-highest Internet user in the world (560,000,000 people), India’s consumption of social media as a bid to be digitally empowered has been anticlimactic where fake news is concerned.

The situation is not too different in the world’s leading economy and the beacon of progress: the United States of America. The backdrop of democratic elections in the USA has been rife with hyper-partisan views due to fake news, playing to the fears and prejudices of the citizens. It is widely believed that people’s voting plans and behaviour were severely altered in the 2016 United States presidential election.

Needless to say, fabricated content and its persuasive nature can be a real threat to society and needs a strong nemesis that can fight it persistently.

Creation of fake news

There was a time when journalists and media outlets used to be the primary sources of information, and they had to verify sources and information before publishing them. But in the bid to provide news stories 24*7, this is not the case now. The propaganda mills work overtime to fill the systems with credible sounding but inauthentic content.

How are fake political, cultural and scientific news stories being created in such large proportions? Unfortunately, Artificial Intelligence (AI) has a large role to play in this. AI tools are being used to populate articles with a credible tone to spur virality. In mid-2019, an article in the BBC (Link to: https://tinyurl.com/yxfsqzwe) reported about a text-generator, built by research firm OpenAI, that could easily create fake news and abusive spam on social media.

Their model, GPT-2, created on the basis of eight million web pages, was able to replicate the style and tone of the initial text but was deemed unsafe, due to the possibility of its malicious applications. The model was later released with fewer parameters but remained controversial due to lack of human intervention and crucial fact-checks. 

The creation of deepfakes (a term coined by a Reddit user in 2018), which refers to doctored videos and photos that can superimpose the physique and facial features of a person, is another version of fake news that can have drastic consequences. From spreading propaganda in order to incite hatred and violence to maligning public figures, deepfakes have been infamous in the news since 2018. AI-powered software can create, enhance and alter a slew of visual data in mere minutes to create detailed imagery with varied consequences.

Disseminators

People are naturally drawn to click-baiting headlines and titillating content. Given the easy access, social media is the largest disseminator of false content. Unwilling allies to fake news, Facebook and Whatsapp have become the two social media platforms that dominate the act of disseminating false content in the world. There are over 2.4 billion active Facebook users worldwide (about 336 million in India).

The statistics for active WhatsApp users stands at 1.6 billion in the world (about 300 million in India). Despite fact-checking teams and actions like limiting group messages, there is a lot left to do to eliminate fake news.

One of the latest measures taken by Facebook to curtail fake news, other than ramping up fact-checking, is its collaboration with the World Health Organization (WHO), to combat COVID-related misinformation. Facebook is providing the World Health Organization free advertising so that it can spread authentic information to the wide base of its users. Among the many updates regarding this topic, Facebook CEO Mark Zuckerberg also announced that the company is working with national health ministries and global organisations to release timely and accurate information on the virus.

Is AI the answer to fight fake news?

AI can be a double-edged sword. While it is actively used to create news, AI intervention can also be helpful in identifying and eliminating it. AI is now looked upon as the cornerstone to separate the fake from the authentic in the news domain. This is because AI makes it easy to learn behaviours through pattern recognition. In the last few years, numerous algorithms have developed to distinguish the difference between human and AI-generated content.

Their primary function is falsehood detection backed by neural network technology, which can highlight discrepancies in articles and determine credibility. Algorithms are developed by feeding them existing articles from various fake news data sets, made of huge virtual data libraries that contain authentic information and sources. Theoretically, AI can scour web pages, weigh facts, predict the reputation of the source by considering aspects such as domain name and Alexa web rank and discover sensational words to detect fake news.

AI-powered analytical tools can also include stance classification to determine whether a headline is in tandem with the article body by processing the text to analyse the author’s writing style and using image forensics to detect Photoshop modifications. This can help highlight anomalies and identify fake news and doctored photographs.

Popular tools to identify and combat fake news

Several companies have tried to combat fake news over the last few years. In the latest efforts to fight COVID-19, Facebook, Google, Twitter and YouTube have come together to limit and eliminate misinformation regarding  the coronavirus pandemic and push official guidelines on their platforms.

Even before the urgency of authentic information in light of COVID-19, companies have been working towards curtailing the systematic spread of inaccurate and misleading information on the Internet and social media. India-based fact-checking platform, MetaFact uses AI to detect and monitor fake news in real time, sifting through all the data that is generated online.

“MetaFact is building a self-sustainable, user-friendly ecosystem to fight fake news, propaganda and misinformation through the use of Artificial Intelligence, Natural Language Processing (NLP)-powered metadata and a conversation-based information distribution platform. Learn more how social media and other industries are using artificial intelligence.

We aim to build social media awareness across communities and help educate the next generation of professionals through the use of digital media literacy programs with a strong focus on the developing world,“ says the website, which displays its fact-checked stories prominently. Other platforms like BOOM and Check4Spam operate in the same domain, verifying social media posts to check stories and tags that can spread misinformation.

Globally, companies like Spike, Hoaxy and CrowdTangle also monitor content, covering colossal amounts of data to check for anomalies. They even predict what is going to drive more engagement and monitor it closely. A company called Pheme has made a technology leap to read the veracity of user-generated and online content. Content and technology giant Google Trends proves its worth by observing searches and picking abnormalities. There are a number of other tools that have found resonance with broadcasters to offer fake-free content.

Popular AI and ML Blogs & Free Courses

Future of fake news

Assessing the legitimacy of news when it is ‘machine-generated’ by AI is only the first step towards eliminating it. Since tools such as auto-completion, text summarisation, etc. are widely used by both credible broadcasters and publishers, there needs to be more checks in place to identify and remove fake news.

Experts continue to be hopeful about the future where each piece of content will be verified by AI technologies in real time before dissemination. The increasing awareness about fake news is another reason why they feel that people will resist information that is inauthentic in nature. With a combination of more sophisticated AI tools and awareness, hopefully there will be less fake news impacting the world.