SIEPR researchers examine exposure to racial diversity
Artificial intelligence has come a long way in a short time. No longer relegated to only outwitting humans in chess matches, AI now powers virtual assistants like Siri and self-driving cars testing their way through our neighborhoods.
But if AI is getting more ingrained in society, why has it not boosted economic growth — as technological innovations like electricity or computers have done in the past?
Erik Brynjolfsson, a leading economist in AI, says it’s only a matter of time.
Brynjolfsson, Director of the MIT Initiative on the Digital Economy, spoke Tuesday at the Stanford Institute for Economic Policy Research (SIEPR), on “The AI Awakening and the Coming Productivity Boom.” The event was co-hosted by SIEPR and the Stanford Institute for Human-Centered Artificial Intelligence (HAI), where Brynjolfsson is a distinguished fellow.
The paradox of our era is that people are less confident about our future than ever, he said. Such concerns are not surprising given how incomes have stagnated, and the rate of productivity — a component of economic output and measure of innovation — has been stuck at about half of what it used to be from over a decade ago.
“We're trying to reconcile some of the amazing things that are happening in technology — thanks to a 50-mile radius of where we are standing now — and why we’re not seeing it in productivity statistics,” Brynjolfsson said.
But a boom is coming, he said. Brynjolfsson points to the recurring lag in productivity growth for other breakthrough innovations — namely, the steam engine, electricity, the internal combustion engine, and computers.
Brynjolfsson thinks AI holds the same promise as what Stanford economist Tim Bresnahan coined as a “general purpose technology” — a major innovation that spawns other advances.
“The technology is real, but the benefits take time to emerge,” he said.
Brynjolfsson used to agree with other economists who attribute the current state of slow economic growth to partly a problem of mismeasurement — where intangible products like search engines and online transactions are not fully captured in balance sheets or the GDP.
But he says the findings of he and his colleagues have since shifted his perspective. He now places more weight behind the “restructuring lag” factor, as investments in AI are still weaving their way across manufacturing, service jobs, and multiple industries.
“When you have these powerful technologies, you need to make a lot of complementary innovations to get their full potential,” he said. “It can take decades.”
Meanwhile, “there’s a gold rush right now” for companies trying to harness the power of AI.
An example: $80 billion was spent on the technology of self-driving cars between 2014 and 2017. And the number of robot chauffeurs today? Zero.
Yet even if digital progress grows the economic pie, it doesn’t mean that everyone, or most people, will benefit, Brynjolfsson cautioned.
There’s no guarantee of shared prosperity: incomes haven’t kept pace with economic growth, he explained. So, “the new grand challenge” will be in reinventing job skills, organizations and metrics to keep up with the accelerating technology, he said.
Susan Athey, a SIEPR senior fellow and professor at the Stanford Graduate School of Business, joined Brynjolfsson on stage for a discussion following his remarks.
“Are you a pessimist or an optimist after all this?” she asked.
“I like to call myself a mindful optimist,” he said. “The more powerful our tools are, the more we can change the world.”