AI and automation are creating a hybrid workforce

• "We've seen more changes in how we work over the past 20 weeks than we have over the past 20 years," says Erik Brynjolfsson, the director of the Digital Economy Lab at the Stanford Institute for Human-Centered AI (HAI).

State of play: The early weeks of the pandemic featured a rush of reports of companies pushing automation and AI, in part to replace humans who couldn't come into the workplace for fear of COVID-19. But the robot jobpocalypse isn't quite here yet, notes Brynjolfsson .

• In a report for MIT Sloan Management Review, Brynjolfsson and his colleague Matt Beane found that while companies are investing in robotics and automation, it takes more time than you might expect for employers to find the right uses for new technology, especially in the midst of an economic downturn when firms are "focused on keeping the lights on," says Brynjolfsson.
• Right now companies are in the early stages of what Brynjolfsson refers to as the "productivity J-curve," often seen when radical new technologies are introduced into the workplace.
• Rather than instantly improving productivity, there's often an initial dip as companies spend time and money to figure out how to get the most out of new tools.

What they're saying: "The reason I don't foresee mass unemployment is that we are still far from artificial general intelligence," says Brynjolfsson. "Machines just can't do soup to nuts what humans can do."

Yes, but: Just because the robots aren't taking all of our jobs doesn't mean that automation and AI won't reshape how we work, even in the short term.

• At a summit HAI held on AI and the future of work earlier this week, McKinsey Global Institute chairman James Manyika estimated that while only about 10% of jobs are at risk of being lost because of automation and AI, 60% of all jobs fall into a category where at least one-third of tasks could be automated.
• "More jobs will be changed than lost," said Manyika. "But this comes with massive polarization and a general sense that work is becoming more fragile."

Context: That polarization can be seen in the fact that roughly two-thirds of the job growth since the 2008 recession has been concentrated in one-third of U.S. counties, Manyika noted, while the rest of the country has experienced either muted growth or declines.

• As my Axios colleague Ina Fried reported yesterday, tech companies are pulling away during the pandemic, with Apple reporting its best September quarter ever in the midst of general economic carnage.

But even the most successful tech companies haven't yet shown that they can't yet replace human workers with robots or AI on a wide scale.

• Amazon has invested heavily in automation in its warehouses, yet this week the company reported it now employs more than 1 million people, with 400,000 jobs added this year alone, mostly in warehouses and delivery centers.

The catch: Many of those jobs fall into the fragile class Manyika described, and as robotics technology improves over time, more and more tasks that can only be done by humans are at risk of being automated away.

• A recent report from the World Economic Forum projected that by 2025, the time spent on tasks by humans and machines will be equal, and that 43% of businesses surveyed said they plan to reduce their workforce due to technology integration.

What's next: Brynjolfsson doesn't want to slow down the pace of technological change, but he argues changes to the tax code — which currently taxes companies more for employing human workers than capital investments in automation — would help cushion the shift.

• He also argues the U.S. needs to invest far more in training workers — including on the job — so they can get the most out of technology.

The bottom line: An essential fact about robots and AI is this: they're always improving. If workers don't learn new skills, says Brynjolfsson, "they'll face some tough times ahead."