Economic winter has been a long time coming; the US has had the longest-running bull market in the country’s history. Markets, startups, and corporate innovation and expansion have flourished for over a decade. But now we appear to be entering the early days of winter, as markets fall on coronavirus threats and large companies—including automotive, retail, and technology firms in the U.S. and Europe—announce substantial layoffs.
Even before the coronavirus-related market declines, my conversations with companies and their consultants indicated that more companies are talking about restructurings and capital expense reductions than expansions. Several economists have raised the odds of a recession in 2020 to 50% or above. It’s too early to call a recession, and I hope, of course, that this economic retrenchment will be short-lived. However, it is clearly wise for any company to prepare a response to dark economic clouds on the horizon.
Which brings up the fate of artificial intelligence (AI) projects in an economic downturn. I don’t anticipate an “AI winter”—a major reduction in AI startups and innovation—to match an economic winter. AI is the most important new technology of the past decade, and it’s an extension of the rise in data and analytics that’s been going on for even longer. The amount of data and the need to make sense of it with analytics and AI will continue to grow. However, I do anticipate a change in focus for AI over the next several years, and I’m already beginning to see it play out.
A Shift in AI Focus
Smartsheet Stock Near Record High After Rallying On Strong Earnings Report
Google Meet Is Now Free For Everyone
One-Third Of Cloud Spending Wasted, But Still Accelerates
The primary focus of AI in many companies thus far has been innovation and exploration—almost to a fault. Firms wanted to learn about the technology and its fit with their strategies and processes, so they initiated many pilots and proofs of concept. Relatively few of these have become production deployments of AI—which means that they haven’t delivered significant economic value. In terms of objectives for AI, in one large company survey which I helped analyze, many respondents reported innovation-oriented objectives like “enhance current products” (44%, the most common response), “free workers to be more creative” (31%), “create new products” (27%), “capture and apply scarce knowledge” (27%), and “pursue new markets” (24%).
The change in economic climate, however, is likely to bring a decreased level of growth in AI investments, and a change in their focus. A January 2020 survey of large U.S. companies by NewVantage Partners—when the economy was still humming along—found a major change from the 2019 survey. When asked if the pace of investment in AI and big data was accelerating, 92% agreed in 2019, but only 52% agreed in 2020.
In terms of objectives, it seems likely that internal and external operational improvements will become a stronger focus than innovation-oriented projects. And labor automation is likely to increase as well as a focus. In the 2017 and 2018 “State of Cognitive” surveys, “reduce headcount through automation” was the lowest-ranked objective, at 22% and 24%, respectively. And the companies I have consulted for or researched have almost always said that they plan to redeploy any workers freed up by automation into other tasks and jobs. I know of very few jobs that have been lost to AI and automation technologies thus far.
But even in the 2018 survey, respondents agreed that “To cut costs, my company wants to automate as many jobs as possible with cognitive/AI.” That sentiment, thus far the focus only of anonymous surveys and back room conversations, is likely to become more prominent in an economic climate characterized by retrenchment. It is a challenging issue for leaders, however; they often require the cooperation and task knowledge of their front-line workers in order to successfully implement automation solutions. But those may be withheld if employees suspect that their jobs are at risk. This set of concerns is most likely to arise with regard to large-scale projects involving automation-oriented technologies like robotic process automation (RPA).
The Return of Technology-Enabled Reengineering?
In the early 1990s, companies were faced with a recession and a challenge from global competitors, particularly in Japan. At that time I published an article and later a book arguing that firms needed to redesign end-to-end business processes (“order to cash” or “procure to pay,” for example) using information technology, with the goal of radical improvement. Other writers—a little later, I’m proud to say—made similar arguments, and the business process reengineering movement came to characterize the first half of the 90s decade. The innovative technologies of that period were enterprise resource planning (ERP) systems and, somewhat later, the Internet.
I’m seeing similar conditions in the early 2020s. The economy is challenging, competition from abroad (notably China) is growing rapidly, and there are new technologies that can drive new ways of doing business. The primary technology enabling change today is AI, including RPA and other automation technologies, as well as process mining to reveal how work is being done.
In a new round of process reengineering, companies could identify—with help from process mining—the end-to-end processes that need the most help, document and measure their current flows and performance, consider how various forms of AI might yield a better-performing process, and determine a new mix of human and machine-based tasks. Those steps could also be undertaken with regard to smaller processes using less dramatic process improvement approaches like Six Sigma and tools like RPA. Indeed, several companies I’ve encountered, including Voya Financial and Lloyds Banking Group, are combining RPA capabilities and process improvement, and ensuring that processes are improved before they are automated.
It’s important that companies don’t repeat many of the mistakes of the earlier generation of reengineering projects. They shouldn’t use the reengineering term—or refer to AI either—to describe layoffs that have no process redesign and no technology enablement. Layoffs that aren’t carefully considered often result in skilled people leaving the company, few increases in efficiency, and jobs that come back when the economy improves.
The Mandate for AI-Driven Productivity
I hope that AI will not become a driver of substantial job loss; for the most part, it’s a better tool for augmenting human capabilities than for replacing them. But neither AI nor any other recent technology has done much to improve national-level productivity in industrialized countries. With population growth leveling off in many of those countries, there is no way to improve economic performance without converting AI capabilities into more output per worker. Some of that increased productivity will probably come from automating tasks and jobs and reducing headcount, though it will take careful analysis and substantial time to do that well.
In other words, it’s a bad idea to wait until a full-fledged recession to begin considering how to lower costs and improve productivity with AI. Real value will come only from production deployments of AI and integration with processes and existing systems. Companies need to begin work now on developing AI applications that create economic value and that lead to new ways of orchestrating work by humans and machines.
*This article was originally published by Forbes on March 12, 2020Share This!