No matter the outcome of the U.S. presidential election, 2024 is likely to mark a new low in political messaging thanks to now-booming generative artificial intelligence technologies.
Chicanery has always been part of politics but never has it had this sophistication or reach. Already, AI has been used to mimic incumbent Joe Biden’s voice to try to suppress primary voting. As the months go by, experts are voicing concern Americans will see and hear swarms of plausible things that haven’t actually happened. Think: convincing fake or altered videos of a downtown riot or an oil spill, a frail Mr. Biden stumbling badly at a campaign stop, or Donald Trump’s physical features altered just slightly to look villainous.
It will all be done to reinforce pre-existing political narratives, or to feed into disinformation campaigns.
“I expect a tsunami,” Oren Etzioni, professor emeritus at the University of Washington, told the Associated Press. “I hope to be proven wrong. But the ingredients are there, and I am completely terrified.”
Terrified is also what one might feel after listening to some of the witnesses at the Standing Committee on Industry and Technology, which is examining Canada’s first-ever legislation on artificial intelligence.
Creating a law on AI appears an almost impossible task. AI has the power to transform society, politics and biology. It is already far smarter and cheaper than researchers predicted a couple of years ago. As both big corporations and individuals harness AI, with various motives, it’s unclear how much governments will be able to control what happens next.
Even still, it’s hard to overstate how crucial it is that Canada build some kind of coherent framework. “Canada is not a big power like China, the U.S. and the EU, of course, but it could – and should – be a contributive player to ensure responsible AI at a global scale,” said Catherine Régis, a law professor at the University of Montreal and a leading voice on AI. The country boasts a large number of leaders in the field, and the country wants to be a part of the global billions of dollars being invested.
At the committee stage, a key question MPs and witnesses have debated is whether it’s best to try to iron out all the flaws of its artificial intelligence legislation, Bill C-27, or pass something quickly with flexibility in regulations that can be changed as technology evolves. Industry Minister François-Philippe Champagne has said failing to act now would mean AI systems already being used across many sectors would remain unregulated for several more years. “Once AI technology already permeates our society, it will be difficult to change expectations or retrospectively address harms.”
You don’t have to look as far as cyberattacks, novel bioweapons or a Terminator-like scenario – or even the future loss of human jobs to AI – to be worried about harms. In the near term, it’s the ability to create convincing images and sounds that will be the most damaging. And it’s not only those in the stratosphere of fame held by presidential candidates, or Taylor Swift (whose likeness was used to create AI-generated porn last month) who will have to face the fakes.
The standing committee heard, for instance, about a young Canadian actor, a minor, whose voice was uploaded to a site that allows users to manipulate it to say “crude, R-rated things,” said Eleanor Noble, the national president of the performers’ union ACTRA.
At the same time, the positives of AI – including advances in medical treatment, and the better management of complicated agriculture and energy systems – are ever-present, too. The wealth and productivity it will create for some is tantalizing. Generative AI semiconductor chip producer Nvidia Corp. – not a household name, yet – is challenging both Google-parent Alphabet and Amazon to be the most valuable U.S. company in existence.
At the committee meeting, Canadian executives from Meta, Google and Amazon expressed concern that Bill C-27 as it stands could be too restrictive, and say the country risks falling behind others.
But in the race for command of AI, many experts are urging caution and constraint, including Yoshua Bengio, a world-leading AI researcher. The University of Montreal professor said even he only truly grasped the magnitude of the risks of “unbridled advances in AI” – ones that now keep him awake at night – after studying the proficiency of ChatGPT last year.
When he appeared at the standing committee this month, Dr. Bengio was clear he believes some guardrails are better than nothing at all. “The current AI trajectory poses serious risks of major societal harms.”
AI will be a problem for the U.S. during this high-stakes election year. But soon, any day now, this will be a problem for us all.
Kelly Cryderman