Canada could boost productivity and GDP using generative artificial intelligence: report

Generative AI technology could reverse Canada’s decades-long decline in productivity and add almost two per cent to the nation’s GDP, according to a report by The Conference Board of Canada.

The report, done in partnership with MaRS Discovery District, recommends that the federal government and organizations work collaboratively to establish guidelines that foster smart regulation and safe adoption of generative AI by Canadian businesses.

Ottawa should focus additional resources on compute power and devise a national strategy to support development of supercomputing and data infrastructure, says the report, “Real Talk: How Generative AI Could Close Canada’s Productivity Gap and Reshape the Workplace – Lessons From the Innovation Economy.”

“I’m not suggesting that generative AI is the panacea. But it certainly is an area where we have a great opportunity to change the course of Canada right now by doing this,” Alain Francq (photo at left), director, innovation and technology at The Conference Board of Canada (CBoC), told Research Money.

“To take advantage of generative AI’s potential, Canadian businesses need to break out of their historic pattern of under-investing in research and development and adopting emerging technologies in an effort to rectify the country’s productivity gap,” he said.

Only six per cent of 16,000 Canadian businesses polled by Statistics Canada reported that they were planning to adopt generative AI over the following 12 months, the report notes. As Research Money reported last December, for companies with 10 or more employees, Canada ranked No. 20 out of 35 countries for AI adoption.

By last April, when generative AI had entered the mainstream, KPMG found that 37 per cent of Canadian companies were exploring ways to use ChatGPT, one of several generative AI programs now available. However, in the U.S. that figure was 65 per cent.

Generative AI describes algorithms (such as ChatGPT) capable of generating new content, including text, audio, code, images, simulations and video.

The CBoC report included an online survey distributed to approximately 1,300 startup companies. A total of 221startups participated in the survey that were operating in 19 sectors and were headquartered in eight provinces. Researchers also conducted 17 in-depth interviews with company founders and experts.

The report found that startup companies have been quicker to embrace generative AI than Canadian businesses at large. Of the 221 companies surveyed 46 per cent were already using generative AI, 38 per cent said they were exploring it, and most said it had boosted productivity.

Four categories emerged as areas of focus for using generative AI: research and development; sales and marketing; operations and production; and customer service and support.

Scientific and technical services is the main Canadian industry using generative AI, followed by healthcare and social assistance, according to the report.

However, AI adoption is particularly difficult for small and medium-sized Canadian businesses, the report says. A study from Toronto Metropolitan University found that 20 per cent of large firms are using AI, but only three per cent of smaller businesses are. The report found that even tech-savvy start-ups are spending significant time and resources in determining where to use generative AI and how to integrate it.

The report recommends that government use financial incentives to de-risk the adoption of generative AI for SMEs. Organizations that support SMEs should consider developing education and awareness programs and creating networks, clusters, toolkits, and best-practice guides to help speed up implementation, the report says.

Canada lacking in compute capacity

When it comes to the compute capacity and associated data infrastructure required to create and use generative AI, the report notes that on the 2023 Global Artificial Intelligence Index compiled by media firm Tortoise, Canada ranked 23^rd among countries for its AI infrastructure.

That rank was eight places below the previous year’s and significantly lower than our performance in other parameters, such as talent and research, in which we were in the top 10, Francq noted.

Canada’s comparative lack of supercomputers and other infrastructure particularly impacts researchers and SMEs, which can be priced out by larger organizations with deeper pockets, the report says. “To reverse this trend, the government will need to consider developing a strategy ensuring that we have the infrastructure in place to be AI leaders.”

Francq said Canada made early investments under the Pan-Canadian AI Compute Environment operating plan, which is part of the Pan-Canadian Artificial Intelligence Strategy. That includes $40 million allotted over five years for the Digital Research Alliance of Canada, which coordinates and funds activities for Canada's national advanced research computing platform  from Compute Canada.

But that’s a small investment compared with other countries, he said. For example, the U.K. is investing £500 million (Cdn$850 million) in its compute capacity. France also is investing heavily in infrastructure and has made its supercomputers available to both researchers and SMEs founded in France and those that choose to relocate there.

Canada needs to build upon that initial funding for the Digital Research Alliance to bolster affordability of compute power and access for the broader population of businesses, including SMEs, Francq said. “That’s actually where we’re going to start to adopt the technology more widely and have that impact on productivity that we’re looking for.”

Barriers to Canadian businesses adopting generative AI

Canada has a highly educated workforce and strong research capability, but consistently lacks commercial success and innovation-based economic growth – a problem known as Canada’s “innovation paradox,” the report points out.

In 1981, Canada’s GDP per capita was about $3,000 above the average of 19 similar OECD countries (adjusted for purchasing power parity. Today, it has fallen to $5,000 below the average. This manifests as low business expenditures in R&D, lagging tech adoption, and a lack of high-tech exports, the report says.

It warns: “If Canada’s productivity continues to be outpaced, the country will be $18,000 behind the OECD average by 2060, resulting in a significantly lower quality of life.”

Generative AI could help boost the productivity of the Canadian workforce, the report says. Google estimates that generative AI will save the average Canadian worker 100 hours per year.

In a study of 453 college-educated professionals, MIT researchers found that those with access to ChatGPT completed a series of exercises – writing cover letters, composing emails, devising business plans – 40 per cent faster than those who had no help from AI.

The non-profit Forum IA Québec predicted that a significant increase in AI adoption could boost Quebec’s GDP by seven per cent to 15 per cent by 2035, the report says.

Yet Canadian businesses lag in adopting and using generative AI. For example, Toronto-based Cohere helps businesses implement AI large language models like ChatGPT. But Ronak Shah, Cohere’s regulatory affairs counsel, said only one per cent or two per cent of Cohere’s customers are from Canada.  

“Canada is a leader in the discovery and creation of artificial intelligence, but we lag in adopting the technology at the organizational level,” Francq said.

The report’s survey revealed several barriers to generative AI adoption by businesses, including high costs, a lack of reliable training materials, concerns over data privacy and security, and uncertainty about how the technology will be regulated.

More than 90 per cent of survey respondents said they weren’t sure about AI rules under current regulations. “They want them [regulations] to be simultaneously loose but tight at the same time, so that you’re able to encourage innovation but also to have those safety guards in place,” Francq said.

Canada needs to “regulate smartly”  when it comes to generative AI, the report recommends. It says federal regulations and guidelines that encourage employers to safely adopt generative AI without stifling innovation are needed. Also, corporations must engage in dialogue with the government around smart regulation for the safe adoption of generative AI in workplaces.

An environment with supportive structures and incentives – including to de-risk experiments in adopting and using generative-AI is needed, Francq said. However, he added that it’s the businesses that are “the actors that have to take the risk and move forward.”

Preparing Canada’s future workforce to use generative AI

Despite Canada’s pool of AI research talent, more than half of survey respondents who are using or exploring generative AI said a lack of technical expertise was holding them back from integrating the technology into their businesses. At the same time, only 29 per cent of those firms said that finding talent was a roadblock to adopting generative AI.

The CBoC report recommends upskilling employees to be able to use generative AI. It’s not yet clear how generative AI will alter the broader labour market, but it seems the most valuable workers will be those who successfully use generative AI to augment their own output, the report says. “Business leaders need to support employees along their learning journey and foster best practices.”

Also, the report recommends preparing the next generation of workers to use generative AI. “Canada’s deep pools of AI talent are a major asset and need to be nurtured further,” it says. “Education sectors need to help prepare students for the changing skill sets required by business and industry.”

Sixty-three per cent of respondents to the survey said implementing AI would not reduce their number of employees. Thirty per cent said they weren’t sure, and a few outliers predicted using the technology would decrease the size of their workforce

It’s not necessarily that AI will replace the worker or the job, Francq said. “It’s the worker that has the ability and the skills to apply AI who will replace someone who doesn’t know how to use those skills.”

The CBoC will be embarking on a large study, collaborating with Vector Institute and other organizations, to benchmarks what is and what isn’t an AI job, he said. “We’re really going to put our finger on the pulse of what those skills are, whether they’re truly technical skills or whether AI competencies and skills will actually show up on the jobs [in the future workforce].”

Canada already ranks “dead-last” among OECD countries for long-term capital growth, “all the way out to 2060,” Francq noted. If Canada fails to move the needle on productivity, he added, “you have to imagine a future where our children are worse off than their parents, or we can’t afford to maintain our social programs.”

Said Francq: “It really is a uniquely Canadian opportunity to adopt generative-AI [to improve the nation’s productivity].”

R$