Trust — or the lack of it — in the age of AI
Every night, as I doomscroll past videos of stunts, accidents, or cute animals, my first reaction is no longer enjoyment or awe or shock. It’s suspicion. Before I reshare anything, I zoom in to look for artefacts: an extra finger here, a weird contortion there, some fault in the lighting or shadows. By the time I’m done “investigating”, I’m either annoyed at seeing yet another covert AI video, or frustrated at not being able to prove what I instinctively feel is fake — even when it sometimes isn’t.
In Nineteen Eighty-Four, British author and journalist George Orwell wrote, “The Party told you to reject the evidence of your eyes and ears. It was their final, most essential command,” imagining it as the logical endpoint of a totalitarian surveillance state. Yet, 40 years after the events of Orwell’s fictional dystopia, “evidence of your eyes and ears” has lost its currency.
In a world drowning in synthetic media, how do we, especially journalists who work with disinformation, trust what we see or hear anymore?
The famous uncanny-valley video of Will Smith devouring a bowl of spaghetti is a relic of the past now — even if it’s a little more than three years old. We now have innumerable text-to-video generative AI tools that create videos so realistic, it’s difficult even for seasoned reporters to debunk. Earlier this year, Hany Farid, the “godfather” of digital forensics, famously struggled to verify whether a video of a US missile attack on Iran was real.
And sure, the internet in the age of AI is a sea of easily identifiable slop right now, from reels of vegetables cheating on their vegetable spouses to videos of Kim Jong Un, Vladimir Putin, and Ali Khamenei pranking Donald Trump. But not all slop is easy to detect. While SynthID and other watermarks have made life significantly easier for fact-checkers, the problem is what synthetic media has done to our collective psyche.
During the recent Nepal floods, a video of an elephant rescuing a man ate up the better part of my workday — simply because I now find it hard to believe anything so miraculous could be captured so conveniently on camera. I should add: I failed to conclude whether or not it was real.
This is not what healthy scepticism looks like.
As fact-checkers, we have been trained to treat social media as a generally low-trust environment. But that shouldn’t apply to every post we come across.
This erosion of trust goes beyond social media. In a low-trust society, every interaction becomes combative. And this becomes fertile ground for bad actors. Politicians can simply call anything and everything that shows them in a negative light a deepfake. Scammers take advantage of mistrust in systems. As Klaus Schwab, the founder of the World Economic Forum, wrote earlier this year: No society or institution can stand for long on foundations that are no longer believed.
Comments