NicoElNino - Fotolia

How to identify projects that create AI business value

When AI projects lead with technology, they rarely have a business impact. Instead, business leaders should target projects to make a meaningful improvement in processes.

Much of the hype that once surrounded AI may be dying down, but that doesn't mean AI business value or adoption is waning.

"It's still thriving, even if it's not making as many headlines as it did a few years ago," said Ben Gesing, director and head of trend research at shipping company DHL.

In fact, enterprises are increasingly moving beyond proof-of-concept projects and implementing AI tools at scale. Beyond that, enterprises are funding and supporting AI projects with specific business value add.

Look for AI projects with broad value

Speaking at the MIT Technology Review's EmTech Digital 2020 conference, which was held virtually for the first time this year, Gesing said DHL is now using AI in its shipping operations in several ways.

In one case, the company is using computer vision tools to inspect shipping pallets to determine how stackable they are. This helps to better apportion space when loading items. It is also using drones to inspect equipment for clients in the oil drilling business that need to move rigs from site to site. By using drones and computer vision deep learning models, DHL can automate the equipment evaluation process and plan how much shipping capacity is needed.

Without AI business value, projects fail

These AI implementations have a direct operational impact, something other adopters said is a crucial component for AI projects.

"If there is no value for the user or the company, you won't get funded," said Luca Finelli, vice president and head of data science and AI insights, strategy and design at pharmaceutical company Novartis.

During a presentation at the conference, he said the relationship between funding and user value became particularly clear after talking with the company's manufacturing team. They weren't interested in a broad effort to digitize their operations. The team would only support projects that led to clear improvement in two performance indicators they measured against: time to production and quality.

"This was a clear experience for us," Finelli said. "We couldn't be driven by a desire to go digital. These guys made it clear to us that if you don't add value, you don't get approved."

AI reshapes old processes to increase value

Enterprises in most industries still have legacy processes that have yet to be transformed by AI. Rather than scrap these processes entirely and start with a new process built around AI, some organizations are adding AI on top of what is already in place.

That is the case at JPMorgan Chase & Co. Speaking at the conference, Apoorv Saxena, the company's global head of AI, said he's focused on implementing AI in ways that provide immediate business benefit, rather than looking to digitally transform a process from end to end.

For example, Saxena said JPMorgan Chase has previously directed massive investments toward the problem of fraud detection. The company has built some fairly sophisticated detection systems that don't rely primarily on machine learning. It wouldn't make sense to ditch those efforts in favor of a new, unproven AI tool, so Saxena is working on adding some learning components to these existing tools to make them more accurate, thereby adding AI business value.

"You have to be very targeted," he said. "You have to bring in something that differentiates and creates a value proposition. It's more about identifying key areas that really move the needle for the business."

Follow the money

What we find is [that] the greatest potential for AI is where the most value can be created.
Michael Chui, partnerMcKinsey Global Institute

When trying to identify projects that would be good candidates for an AI transformation, just follow the money, said Michael Chui, a partner at the McKinsey Global Institute.

Speaking at the conference, Chui said the majority of AI-related headlines people see are about transformative business models built on deep learning or super-advanced AI applications. But most businesses are finding success with more narrow applications of AI to their biggest cost drivers, like supply chain or product development, and their largest potential revenue generators, like marketing.

"While it's easy to be misled by the headlines, what we find is [that] the greatest potential for AI is where the most value can be created," Chui said.

Dig Deeper on AI business strategies

Business Analytics
Data Management