AI and the Fake News War
In a “post-truth” era, AI is one of the many protective tools and weapons involved in the battles that male up the current, ongoing “fake news” war.
Fake news has become widespread in recent years, most prominently with the UK Brexit referendum, the 2017 UK general election, and the U.S. presidential election, all of which suffered interference in the form of so-called ‘fake news’ / misinformation spread via Facebook which appears to have affected the outcomes by influencing voters. The Cambridge Analytica scandal, where over 50 million Facebook profiles were illegally shared and harvested to build a software program to generate personalised political adverts led to Facebook’s Mark Zuckerberg appearing before the U.S. Congress to discuss how Facebook is tackling false reports. A video that was shared via Facebook, for example (which had 4 million views before being taken down), falsely suggested that smart meters emit radiation levels that are harmful to health. The information in the video was believed by many even though it was false.
The Digital, Culture, Media and Sport Committee has published a report (in February) on Disinformation and ‘fake news’ highlighting how “Democracy is at risk from the malicious and relentless targeting of citizens with disinformation and personalised ‘dark adverts’ from unidentifiable sources, delivered through the major social media platforms”. The UK government has, therefore, been calling for a shift in the balance of power between “platforms and people” and for tech companies to adhere to a code of conduct written into law by Parliament and overseen by an independent regulator.
One way that social media companies have sought to tackle the concerns of governments and users is to buy-in fact-checking services to weed out fake news from their platforms. For example, back in January London-based, registered charity ‘Full Fact’ announced that it would be working for Facebook, reviewing stories, images and videos to tackle misinformation that could “damage people’s health or safety or undermine democratic processes”.
A moderator-led response to fake news is one option, but its reliance upon humans means that this approach has faced criticism over its vulnerability to personal biases and perspectives.
Automation and AI
Many now consider automation and AI to be an approach and a technology that are ‘intelligent’, fast, and scalable enough to start to tackle the vast amount of fake news that is being produced and circulated. For example, Google and Microsoft have been using AI to automatically assess the truth of articles. Also, initiatives like the Fake News Challenge (http://www.fakenewschallenge.org/) seeks to explore how AI technologies, particularly machine learning and natural language processing, can be leveraged to combat fake news, and supports the idea that AI technologies hold promise for significantly automating parts of the procedure human fact-checkers use to determine if a story is real or a hoax.
However, the human-written rules underpinning AI, and how AI is ‘trained’ can also lead to bias.
Deepfake videos are an example of how AI can be used to create fake news in the first place. Deepfake videos use deep learning technology and manipulated images of target individuals (found online), often celebrities, politicians, and other well-known people to create an embarrassing or scandalous video. Deepfake audio can also be manipulated in a similar way. Deepfake videos aren’t just used to create fake news sources, but they can also be used by cyber-criminals for extortion.
There has also been a case in March this year, where a group of hackers were able to use AI software to mimic an energy company CEO’s voice in order to steal £201,000.
What Does This Mean For Your Business?
Fake news is a real and growing threat, as has been demonstrated in the use of Facebook to disseminate fake news during the UK referendum, the 2017 UK general election, and the U.S. presidential election. State-sponsored politically targeted campaigns can have a massive influence on an entire economy, whereas other fake news campaigns can affect public attitudes to ideas and people and can lead to many other complex problems.
Moderation and automated AI may both suffer from bias, but at least they are both ways in which fake news can be tackled, to an extent. Through adding fact-checking services, other monitoring, and software-based approaches e.g. through browsers, social media and other tech companies can take responsibility for weeding out and guarding against fake news.
Governments can also help in the fight by putting pressure on social media companies and by collaborating with them to keep the momentum going and to help develop and monitor ways to keep tackling fake news.
That said, it’s still a big problem, no solution is infallible, and all of us as individuals would do well to remember that, especially today, you really can’t believe everything you read and an eye to source and bias of news coupled with a degree of scepticism can often be healthy.