REMEMBER when none of us could agree whether that viral dress was blue and black, or white and gold? It may be a sign of things to come.
A bizarre future where we can no longer trust our eyes might await us as a top AI expert reveals how deepfakes might reshape the future, speaking to The Sun on Future Tech Feed with Sean Keach.
From ‘deepfake’ voice clones that scam your poor gran out of her life savings to hyper-targeted propaganda designed just for you, AI risks rewriting our reality.
Cyber expert Mhairi Aitken, of Our AI Collective, revealed how it’s not just that AI has got better, but also that we can all use it so easily.
“There’s this big leap forward in what generative AI is able to do,” said Mhairi, who previously worked at The Alan Turing Institute.
“But also that these tools became available and accessible to pretty much anybody with internet access.”
DEATH OF THE ‘WONKY HAND’
Until quite recently, you could spot a fake image very easily – just look for six fingers on a hand or an ear in the wrong place.
But those days are long gone, according to Mhairi, who said that the pace of changing is “dramatic”.
Just a few years ago, “deepfakes” of celebs like Tom Cruise were poor quality and could only made by the computer-savvy.
Now anyone with a smartphone can do it for free.
“You no longer need to be a computer programmer. You don’t need to know how to code to be able to use these tools,” Mhairi told The Sun.
“And to very easily be able to create quite convincing outputs, whether it’s images, videos, or audio.”
It used to be that pro Photoshoppers would airbrush celebs – and then it moved to social media filters that let us do it to ourselves.
Now AI is so convincing that anyway can create almost any reality at the touch of a button.
But although that could be a problem for body image, there are far more nefarious uses.
“More than 90% of AI-generated deepfakes are non-consensual pornography,” Mhairi explained.
“And it’s not something that’s just targeted at high-profile figures.
“It’s targeting and affecting everyone, including the most vulnerable members of our society.
“And so this is something that really requires urgent action.”
Mhairi warned that the problem is even in schools, with explicit deepfakes being shared between youngsters “contributing to the existing challenges around bullying”.
‘LIAR’S DIVIDEND’
The problem isn’t just that you might believe something fake.
It’s also that you might not believe something that is actually real.
Mhairi says this problem – known as the “Liar’s Dividend” – is the ultimate get-out-of-jail-free card.
The idea is that if a dodgy politician or criminal is exposed as doing something unscrupulous on camera, they could simply shrug and say it was AI.
“It’s a defence against real incriminating evidence,” Mhairi said.
“Your first response could be well, oh, it might just be AI-generated.
“How do I know that’s real? And I think that’s a much bigger risk. And it’s fundamentally a risk to democracy.”
And this problem becomes worse because of our social media bubbles, Mhairi reckons.
“Increasingly, people are consume information about the world in echo chambers,” she said.
“Where you only come across information that reinforces or reaffirms their existing viewpoints and beliefs.
“That makes it particularly dangerous, because if you’re seeing AI-generated videos or AI-generated images that are depicting inaccurate or false information about the world – but that’s all you’re seeing – you’re not seeing the other side.
“You’re not seeing that being challenged.”
YOUR EYES FOR SALE?
The situation might soon get even stranger as the problem moves from your phone to your face.
Tech giants Meta, which owns Facebook, is pushing smart glasses that now have displays built in.
And Apple has “spatial” headsets that can transform the world before your eyes in stunning quality.
At some point in the next two decades, we might see gadgets that look like regular glasses but with the visual quality of a VR headset.
And that means you might start seeing the world through a digital lens that’s controlled by a billionaire in California.
That might mean you can change the colours of walls at the blink of an eye, or it could mean the glasses control what advertisements you see.
But it could also mean that you get “hyper-personalised propaganda”.
“We might each get a slightly different version of an AI-generated deep fake or a fake news story,” Mhairi said.
“And we all get a slightly different version that is going to align more closely to our interests and our beliefs.”
This can happen easily on social media, but it could be even more damaging when it’s actually your entire reality changing before your eyes.
Mhairi said we need to have conversations about what we should or shouldn’t be allowed to do with AI.
“That can’t be left to big tech companies, to social media platforms,” she said.
“It needs to be a public discussion about: what do we think is socially acceptable? What do we actually think is beneficial?”
