AI Won’t Solve Fake News (Not Today, Anyway)

  4 m, 0 s

Even while Mark Zuckerberg testifies before Congress, social networks are struggling to fight back against fake news. Truth is, AI is really good at creating fake news, but not so good at detecting it. Today we’re exploring why AI won’t solve fake news, and how the problem will only get worse before it gets better.

[what-is-fake-news.png]What is “fake news”?

First, let’s define terms and make sure we’re on the same page. Here’s how the Irish Internet Safety Awareness Centre defines fake news:

“Fake news is news, stories or hoaxes created to deliberately misinform or deceive readers. Usually, these stories are created to either influence people’s views, push a political agenda or cause confusion and can often be a profitable business for online publishers.” – Webwise.ie

The epicenter of the debacle is, of course, Facebook. Mark Zuckerberg & Co. have tried time and time again to fight misinformation on their platform (even as they stumble over accountability and consequences.) But the fake news epidemic reaches far beyond social media platforms.

[ai won't solve fake news.png]AI won’t solve fake news on Facebook (or anywhere else)

For now, let’s set aside questions of moral responsibility, ethics in data science, and Cambridge Analytica. Instead, let’s talk about this recent Verge article, “Why AI isn’t going to solve Facebook’s fake news problem“.

“The company doesn’t seem to have a clear strategy,” James Vincent writes. “Instead, it’s throwing a lot at the wall and seeing what works.” For example, Mark Zuckerberg told the New York Times that Facebook has used “some AI tools” to flag fake accounts and false news. But then Zuckerberg clarified that they’d only assessed suspicious behaviors, not article content itself. That’s a good thing, Vincent says, because “when it comes to fake news, AI isn’t up to the job.”

Why is AI so bad at fighting fake news?

Despite major technological strides, it’s still really difficult to train an artificial intelligence to distinguish between irony, sarcasm, and sincerity. In order to detect fake news, an AI must be able to understand the intent behind a piece of text, an image, or a video. And in the era of “it was just a joke, bro“, the proverbial “sincerity line” gets blurred past recognition.

For example, consider this Youtube thumbnail image:

[tidepodchallenge.jpg]

Image source: The Verge

Based on the thumbnail alone, does the video promote the Tide Pod Challenge, or warn against it? The answer may seem obvious to you, a human. But consider it from an AI’s perspective. Should the computer focus on the image of a person eating a tide pod with a smile on his face, or the text saying “STOP!!!” – and which factor should it weigh more heavily?

This is just one of the challenges an AI system faces in trying to detect fake news. And we haven’t even started discussing accusations of censorship and bias. Even when they work as intended, AI solutions often create entirely new problems.

For now, we’ll leave it to The Verge to explain more.

[tunnel.png]Is there any light at the end of the tunnel?

Well… maybe. It’s clear that AI won’t solve fake news, at least not yet. In fact, artificial intelligence will make things even worse, says Patrick Watson of Mauldin Economics.

Watson points to recent breakthroughs in “fake” videos. In one example, researchers used AI create convincing lip syncs.

The result: President Barack Obama appearing to say fabricated statements.

Frightened yet? Just imagine a falsified video of Warren Buffett telling investors to dump a certain stock. Or a clip of Mark Zuckerberg announcing his candidacy for president. How would the markets react? What about a fake speech of a politician declaring war?

“Because of the latest technological breakthroughs in AI, reality itself – or our perception of it – could be at stake.” – Patrick Watson

So, if AI won’t solve fake news, what can we do? Watson has a few suggestions, such as finding “informed yet skeptical curators who can recognize what’s real and valuable”. Build your “circle of trust”, and favor quality over quantity.

Further reading

The rabbit hole of artificial intelligence and fake news goes very deep. If you’d like to keep reading, we suggest these articles in addition to the ones linked above:

Can AI Win the War Against Fake News?” – MIT Technology Review

The Platform is the Message – Georgetown Law Technology Review

Could AI Be the Future of Fake News and Product Reviews?” – Scientific American

In the meantime, stay tuned to our blog for more AI News & Insights, fresh perspectives and interesting analyses.

And if you’re interested in natural language processing, visit our resources library for a great collection of white papers, data sheets and more.

Categories: Artificial Intelligence, Newsletter