Last week, hundreds of tech leaders signed an open letter calling for a brief moratorium on new developments in artificial intelligence. The letter raises big questions: “Should we automate away all the jobs, including the fulfilling ones? Should we develop non-human minds that might eventually outnumber, outsmart, obsolete and replace us?”
Those are worthwhile questions, and it is refreshing to see some of the people who have helped us get to this point start to ask them. Whether it is even possible to stuff the electric genie back in the hard drive, as it were, the letter is right that there is urgency to an issue that will have profound effects on the economy and society. Which is why Ottawa needs to start taking AI advances seriously.
On the economic front, there is both good and bad news.
The good is that AI promises to boost productivity, a measure on which Canadian firms have long lagged behind. Delayed uptake of automation is one reason – though not the only one – for weak productivity. AI can help.
As part of that, the government should do what it can to invest in the country’s AI industry. An early international leader, it has started to fall behind because of a lack of focus on commercialization and intellectual-property retention.
But the bad news is obvious, too. Robots can take your job. There is a real, visceral fear among workers that their jobs could be automated away, including those of knowledge workers who once thought themselves immune. Ottawa can address this through investments in skills training and modernization of the employment insurance program. Economic research has suggested automation tends to replace rather than remove jobs, and holds the potential for new jobs as firms become more productive. We need to ensure workers have the right skills to take advantage of those opportunities, so the new wealth can be shared.
AI’s potential impact on society is stark, and has already been felt. Social-media feeds, driven by secret algorithms, surface content that makes voters more angry and teens less healthy. There are also emerging potential harms in myriad fields, from self-driving cars to facial recognition.
On this, the Liberal government has begun to take steps, though far too slowly. Last summer, it introduced Bill C-27, one part of which – the Artificial Intelligence and Data Act – seeks to bring some oversight over AI. In one example of that, the bill should clarify who is accountable when something goes wrong.
But the bill has received only a few hours of debate, and no votes. And even when it is passed, it calls for two more years of consultation and regulation-drafting, with new rules coming into effect – at the soonest – sometime in 2025.
That’s unacceptable. Debate on the bill needs to begin in earnest and the timeline moved up significantly. The advances in AI are happening too quickly; ChatGPT, a generative language AI that can, among other things, write software for cyberattacks, debuted in November and already boasts more than 100-million users.
At the same time, the bill needs to be strengthened. It creates an AI and data commissioner to serve as watchdog over the sector. But the commissioner would be answerable to the industry minister and part of Innovation, Science and Economic Development – the same department tasked with promoting the sector. It’s an obvious conflict of interest.
The commissioner needs to be independent – possibly even answerable to Parliament, as is the Privacy Commissioner. (Who, incidentally, announced Tuesday he was probing ChatGPT.) The Globe and Mail recently investigated the Canadian Ombudsperson for Responsible Enterprise, an office created by the Liberal government in 2018 to ensure Canadian businesses behave ethically overseas. The office has been totally ineffective, in part because it lacks any investigative authority. Similar complaints have been levelled at the Competition Bureau, whose commissioner, Matthew Boswell, has called for more powers and funds. An AI watchdog needs the resources and mandate to do its job properly.
Fully blocking new AI technology is unworkable, but letting the sector run wild is not an option. Ottawa has a role to play in crafting rules to allow safe, productivity-boosting uses of AI to flourish, while punishing bad actors.
But time is of the essence. If politicians and public servants can’t act fast or effectively enough on this, well, perhaps we should hire robots to do the job for them.