At this point, it’s still too early to assess whether AI is going to deliver a major boost to economic productivity, says CIBC World Markets Inc.
In a report published Thursday, the firm’s economists call for caution on reading too much into recent economic data from the U.S., which indicated that output grew in recent quarters, without much employment growth.
“Some see that combination as the dawn of a new era of productivity growth, and there’s plenty of hype about the future being even brighter on that front as AI adoption builds,” it said. However, it cautions against believing that hype just yet.
For one, recent data is too unreliable to identify a decisive trend in productivity, the report suggested, particularly given the trade turmoil and other forces at play in the economy.
Moreover, it noted that previous episodes of major technological innovation have historically taken time to gain traction.
“In many cases, new technologies took a lot longer than expected to show up in the productivity data,” the report said.
And, when it comes to boosting productivity, it’s not just a matter of generating more output with less labour, what matters more is total factor productivity, it noted, which factors in both labour and capital inputs.
“Replacing workers with software will only generate an adequate investment return if the capital and electric power inputs required don’t swamp the labour cost savings,” it said.
Indeed, there are a number of hurdles to be overcome before AI can generate large productivity gains, it noted — including high capital costs, chip supplies and the possibility of rising labour costs in sectors that aren’t easy to automate.
Ultimately, it concluded that “it’s simply too soon to have a firm view of what AI will mean for productivity.”
While the technology has potential to boost productivity, that will likely take time, the report said.
“As a result, we’re not building in a sea change in productivity in our forecasts for 2026/27.”