The urgent case for AI regulation
It’s predicted to both revolutionize health care, and if we’re not careful, kill us all.
However, judging by its current use, no one would blame you if you assumed that the primary purpose of artificial intelligence — specifically the chatbot known as ChatGPT — was to generate screenplay ideas or to produce editorials like this one. (For the record, it didn’t).
This rapidly emerging technology promises to dramatically transform elements of the world around us, from office work to commerce to the arts to journalism.
Yet we are just playing with its uses and trying to comprehend its potential. It seems the human race is, at present, basking in the novelty of artificial intelligence without thinking deeply about how this technology will fundamentally change our lives. Currently, many of us use these tools to generate wacky images, write passable high school essays, or produce a fake, albeit very catchy, Drake song. But the honeymoon phase with artificial intelligence will soon come to an end, and a new phase will begin: one in which we stop guessing at the huge impacts AI will have on our society and instead, actually begin to feel those impacts.
Although some experts worry that AI will learn to dominate humanity and eventually destroy it, others argue there are more immediate threats to our civilization than a robot takeover, and these threats are frightening in their own right.
For example, AI has great potential to perpetuate discrimination, damage democracies and eliminate jobs on a mass scale.
According to researchers Dylan Baker and Alex Hanna at the Distributed AI Research Institute in the United States, there are “countless” instances where AI was deployed with “overtly racist, sexist, and discriminatory outcomes.”
It’s no wonder the World Health Organization recently issued a call for “safe and ethical AI … to protect and promote human well-being, human safety, and autonomy, and preserve public health.”
Perhaps more surprising — and alarming — is that even tech leaders are wary of the technology. Earlier this year, Geoffrey Hinton, the so-called Godfather of AI (whose research at the University of Toronto was revolutionary in developing AI technology), recently quit his job at Google, in part, so that he could speak freely about the potential dangers of AI.
Thousands, including Apple co-founder Steve Wozniak and Tesla CEO Elon Musk, recently signed an open letter demanding “all AI labs to immediately pause for at least six months the training of AI systems more powerful than GPT-4.”
Meanwhile, in Canada, dozens of scientists, including Yoshua Bengio, a leading expert in AI, signed an open letter urging MPs to quickly pass, the Artificial Intelligence and Data Act: a law that would regulate AI systems and codify financial penalties for bad actors.
Recently, Sam Altman, the CEO of OpenAI, the company behind ChatGPT, told a panel at the U.S. Senate that he worries about the tool’s ability to manipulate and persuade people and the implications of all that on voting outcomes.
“Given that we’re going to face an election next year and these models are getting better I think this is a significant area of concern,” Altman said. “My worst fear is that we cause significant harm to the world.”
It is a fear we share. World leaders long ago missed the opportunity to properly regulate social media: a failure whose effects are felt in every corner of our society today.
The time is now for governments to come together to regulate AI. The federal government must act to pass the Artificial Intelligence and Data Act, tabled last June under Bill C-27. As critics have argued, the legislation isn’t perfect. It isn’t bold enough, or specific enough. However, when it comes to a rapidly evolving technology, our leaders must get these regulatory laws on the books first and refine them later.
Ultimately though, a piecemeal approach won’t work to reign in a technology that is sweeping the globe. Nations must work together to regulate and monitor these systems — and they must do so immediately. The stakes are far too high for further delay.