Companies Are Rushing to Use AI—but Few See a Payoff

A study finds that only 11 percent of firms that have deployed artificial intelligence are reaping a “sizable” return on their investments.
DHL sorting facility
Photograph: Monika Skolimowska/Getty Images

At some DHL shipping centers, artificial intelligence now helps employees make sure pallets will load safely into cargo planes. A computer vision system captures each pallet, and an algorithm judges whether it can be stacked with other pallets or may be too awkward to fit on the next flight.

DHL is one of a growing number of companies using AI. Besides the pallet scanning system, AI helps route deliveries, control robots that ferry packages around warehouses, and control an experimental robot arm that picks and sorts parcels. DHL is also among a small minority of companies using AI—just 11 percent—that say they’ve reaped a significant return on investment from using the technology, according to a new report.

The report, from Boston Consulting Group and MIT Sloan Management Review, is one of the first to explore whether companies are benefiting from AI. Its sobering finding offers a dose of realism amid recent AI hype. The report also offers some clues as to why some companies are profiting from AI and others appear to be pouring money down the drain.

One key: continued experimenting with AI, even if an initial project doesn’t yield a big payoff. The authors say the most successful companies learn from early uses of AI and adapt their business practices based on the results. Among those that did this most effectively, 73 percent say they see returns on their investments. Companies where employees work closely with AI algorithms—learning from them but also helping to improve them—also fared better, the report found.

article image
Supersmart algorithms won't take all the jobs, But they are learning faster than ever, doing everything from medical diagnostics to serving up ads.

“The people that are really getting value are stepping back and letting the machine tell them what they can do differently,” says Sam Ransbotham, a professor at Boston College who coauthored the report. He says there is no simple formula for seeing a return on investment, but adds that “the gist is not blindly applying” AI to a business’s processes.

AI became a hot business buzzword after research showed how machine learning algorithms could perform certain tasks with superhuman skill—when fed enough training data and computer power. In recent years, it has become clearer that AI often still needs a helping hand from humans to perform well.

The new study surveyed 3,000 managers at companies across a range of industries, as well as executives and academics. More than half of the managers—57 percent—said their company was piloting or using AI, up from 44 percent in 2018.

That’s far more common than suggested by a recent US Census report, which found that relatively few businesses across the entire economy have begun using AI. The BCG report focused on larger companies, most with annual revenue above $100 million. As more businesses adopt AI, more effective use of the technology will provide a competitive edge.

The BCG report classified a sizable return on investment as $100 million in new revenue or cost savings annually for companies with annual revenue of $10 billion or more. For companies with revenue between $500 million and $10 billion, a sizable return was defined as $20 million; and for companies with revenue between $100 million and $500 million, the threshold was $10 million.

The researchers behind the study used machine learning (naturally) to analyze the survey results, and identify key insights from companies seeing a significant return on investment for AI.

The report highlights businesses that implemented AI as part of a bigger rethink of how they operate, and saw greater returns as a result. Repsol, for example, a Spanish energy and utility company, uses AI to identify problems with its drilling operations; to coordinate blending, storage and delivery of oil; and to automatically generate offers for customers. But the report suggests Repsol benefits most from how it learns from these processes, deploying new business practices as a result.

DHL’s use of AI, which is highlighted in the study, also offers insight on why certain companies are benefiting financially from AI when others aren’t. Gina Chung, vice president of innovation at DHL, says the company began incorporating data science, analytics into its business as part of a broad overhaul eight years ago, adding projects related to machine learning about five years ago.

Chung says humans often work closely with AI systems at DHL. Packing pallets onto an aircraft requires experience and skill. An expert loader can train an algorithm to recognize which pallets can be stacked, or how irregularly shaped ones may fit together. This allows the process to be done automatically or without an expert on hand, but the algorithm will make mistakes, especially early on, requiring oversight for some time.

Humans work together with AI systems elsewhere at DHL. A person can, for example, take control of a prototype robot arm if it fails to sort a package correctly. The intervention can be used to retrain the algorithm controlling it. “A lot of these systems powered by AI, they're not 100 percent perfect, especially in the early stages of deployment,” Chung says. “You involve the experts to kind of help improve the accuracy of the algorithm.”

The new report points to other examples of AI-human teamwork, including an unnamed financial company that trained algorithms by studying the behavior of human traders, then has humans learn from the performance of those algorithms.

“We’re seeing that this blending of humans and machines is where companies are performing well,” says Ransbotham. “It’s also that these companies have different ways of combining humans and machines.”

Another example in the report shows the importance of oversight and flexibility in deploying AI. Lyft, the ride-sharing company, developed an AI algorithm to maximize revenue by matching drivers and riders. But data scientists then noticed that the company would get a bigger payoff if it instead focused on maximizing how often users ordered a ride after opening the app. So the first algorithm was scrapped in favor of a different one.

“The idea that either humans or machines are going to be superior, that's the same sort of fallacious thinking,” says Ransbotham.


More Great WIRED Stories