It’s too soon to tell whether AI is making us more productive: ‘People are underestimating the long-term impact of AI at scale’
Is AI going to unleash a new productivity revolution, like electricity? Or will its effects be more modest, like the PC?
Is AI more like electricity, or the PC? To optimists, AI could unleash a wave of productivity gains, much like mass electrification did in the late 19th and early 20th centuries. Last year, Goldman Sachs researchers predicted that AI could add as much as 7% to the world economy by 2033.
Yet not everyone buys into the hype: MIT economist Daron Acemoglu, in an April research paper, argued that AI’s boost to the GDP will be much more modest, at just 1.1% over the next decade. Previous technological advancements, like the PC or the internet, had more limited productivity gains compared to previous “industrial revolutions.”
But speakers at Fortune Brainstorm AI Singapore last week tried to reassure attendees that AI really could lead to productivity improvements—at least, after a while.
“People are underestimating the long-term impact of AI at scale, and overestimating the short-term,” said Panos Madamopoulus-Moraris, founding managing director of industry programs and partnerships at Stanford University’s Human-Centered AI Institute.
Companies could end up having an unrealistic expectation of when they might see a return on their AI investment, he warned. Large corporations will take time to get their AI operations fully online, he said.
Anant Maheshwari, president and CEO for global high growth regions at Honeywell, suggested that aviation could be an industry that could be revolutionized by AI.
He admitted that most people would be uncomfortable taking a fully autonomous plane for a long-haul flight. But AI could augment the capabilities of human pilots, allowing airlines to fly planes with fewer people—and thus allow more flights overall.
“Today, it’s regulation two pilots in the aircraft to fly you. Will it get down to zero? Maybe not. But can it get down to one with the power of AI and therefore have a lot more aviation going? That’s possible,” Masheshwari said.
Inversely, Maheshwari suggested that AI could also help with the opposite problem: Countries dealing with a labor surplus, rather than a labor shortage. Developing countries “need to train up a lot of people for these new industries,” he said. “That’s again where you need AI to help that labor come up the learning curve very quickly.”
Some commentators worry that AI could drive mass unemployment, as white collar workers are automated out of a job. Last year, Goldman Sachs warned that AI could put 300 million full-time jobs at risk.
And companies are already using AI as a reason to lay off workers. In January, United Parcel Service cut 12,000 managers, potentially the largest layoffs in the company’s history. At the time, CEO Carol Tomé said the cuts were possible thanks in part to AI, citing an example where salespeople no longer needed to consult pricing experts to create proposals.
Yet Stanford’s Madamopoulus-Moraris wasn’t too worried about AI entirely replacing humans. He cited the concept of complementarity, saying that humans are better than AI in tasks that require emotional intelligence, creative problem solving and contextual understanding.
And he noted that, for many industries, the problem is having too few workers to do the job, not having a lot of inefficient human workers.
“There was a lot of talk in the healthcare community about radiologists and how radiologists are going to be replaced. Well guess what? The U.S. doesn’t have enough. There are two times the number of postings to people who are actually available to work as radiologists,” Madamopoulus-Moraris said.
