When even Geoffrey Hinton, who is known as the father of Artificial Intelligence (AI) for pioneering 'deep learning' based on neural network that can recognize faces and objects like humans do, leaves Google to warn people about the dangers of this technology, then everyone should listen. He thinks these 'smarter-than-human' generative AI poses anexistential threat to humanity. The generative AI is a type of AI technology that can create different types ofcontents that include text, audio, video and synthetic data. This University of Toronto professor after developing his visual recognition software in collaboration with his students in 2012 is perhaps feeling sad about what he had worked so hard for. He in his recent interview with the New York Times, however, expressed his concern more about the harm that 'generative intelligence' could do through spreading misinformation. His real concern, though, was that digital intelligence is fast outperforming human intelligence! And not only Hinton, more than 27,000 people that included AI researcher and executives of tech companies signed an open letter calling for a halt to 'training the most powerful AI systems' for at least six months.
Since the new AI technology can compose texts, create voices and imageries on its own, people may mistake those for real. Some recent incidents of fraudulence by faking voices of real people (which is about deepfaking) have sent shivers down the spine of many. Some months ago, a mother of two in Arizona of the USA was the target of a deepfake kidnapping scam. Her 15-year-old daughter was sobbing asking for help saying some bad guys were threatening to kidnap her demanding a ransom of USD1 million. When the mother, JenniferDeStefano, phoned her daughter who was out of town on a ski trip, the truth came out. Her daughter was fine. The previous panic-stricken voice imploring her to send help, was also exactly like her daughter's, though it was a complete fake.
It was finally found out to be a hoax, a work of generative AI. Similar incidents have been reported from other places across the globe. AI-generated images of 86-year-old Pope Francis in a stylish puffer jacket and a silver cross hanging down to his chest went viral that stunned millions who viewed it. But it was false. Similarly, a fake song of Drake and The Weeknduploaded in mid-April by an anonymous TikToker was viewed by 11 million people. But these incidents of impersonating people or faking events in the cyberspace was short-lived as those were discovered before they could do serious harm to people targeted. But things may not remain so simple in the near future when the AI tools behind these mischiefs get smarter. There is no denying the fact that cybercriminals have now got a new, dangerous weapon in their hands. People's voice, pictures, their personal histories are all available on the internet. The social media, various phone apps, news media, you name it, provide anendless stream of biometric data to which anyone with criminal intent has easy access. What the new AI technology needs is a tiny clip of the targeted person's audio, a very small piece of his /her video image to impersonate the victim. The person thus faked may suffer huge financial loss or damage to her/his reputation.
Financial institutions and security establishments will get more vulnerable than before, thanks to the deepfaking potential of the latest AI technology. To think that self-guided robot drones operated by terrorists attacking targets with small, tactical nuclear devices!
It is no longer a plot of science fiction. It may happen any time.
So, untilfoolproof AI-tools to counter these deepfakersare in hand, human society is indeed at grave risk. Dr. Geoffrey Hinton has a point.
© 2024 - All Rights with The Financial Express
