AI is increasingly becoming a business imperative. Nine in 10 Fortune 1000 companies are not only investing in AI, but are increasing those investments, with 92 per cent reporting measurable business benefits from their current AI use — up from 72 per cent in 2020 and just 28 per cent in 2018, according to a 2022 NewVantage Partners executive survey.
Still, only 26 per cent of companies say their AI initiatives have actually moved into widespread production. The biggest obstacle? Cultural barriers, with executives 11 times more likely to say culture is the greatest impediment to AI success than to cite technology limitations as the biggest barrier.
And the cultural challenges have actually gotten worse, with 92 per cent of executives citing cultural factors this year vs. 81 per cent in 2018.
The upshot? Companies are finding that the key to successfully operationalizing AI comes down to people, and putting them at the centre of their initiatives.
Putting AI into delivery trucks
When Michael DiMascola, safety business partner at Herr’s Foods, wanted to reduce accidents for its delivery trucks, the first thought was to install surveillance cameras to watch drivers.
The Pennsylvania-based maker of potato chips, cheese curls, and other snacks operates a fleet of 640 vehicles to distribute products in the eastern United States and Canada, and drivers already had a bad taste in their mouths from a previous attempt to install cameras in their cabs.
“The stigma was that Big Brother was watching,” DiMascola says. “And they lit up like a Christmas tree when an event happened, so it was more of a distraction.”
If the problem is that drivers are too distracted, then adding yet another distraction isn’t going to help, he concluded. Plus, the old cameras only triggered after something bad happened, such as a collision or sudden braking or acceleration. “We needed to get ahead of those events,” says DiMascola, who saw distracted driving as a top priority to address.
So this time around, Herr’s took a different approach. DiMascola found potential vendors at a national safety conference in 2018 and started a pilot project with Nauto, a maker of AI software for driver and fleet safety, that fall.
The new cameras DiMascola wanted to deploy paid attention to where drivers are looking so they can alert them if their eyes strayed too long from the road — something that could have potentially been perceived as even more intrusive than the first set of cameras Herr’s used.
“That was something we were very concerned about,” DiMascola says. “Just another Big Brother piece, but now in every truck.”
The rumor mill started up quickly, he says. To get ahead of this, DiMascola needed a personal touch. For the first deployment, he picked two busy locations and trucks that were part of the company’s tractor-trailer fleet. “I went personally to each and every one of our locations, and sat personally with each and every one of our drivers,” he says.
He built a rollout program that included the vendor’s welcome videos, and then showed recordings of collisions, attempted fraud claims, and near misses. “We took our time and made sure that questions were asked, FAQs were out there,” he says. “The training was absolutely the most important piece of it.”
The end goal was to keep the drivers as safe as possible. And taking eyes off the road, even for a short time, can be deadly. “At 50 miles an hour, in 5 seconds you’ve travelled a football field and a half,” he says. “That’s a long distance to be distracted.”
The new platform puts data into the drivers’ hands, he adds. They can see the results in real-time — not just instances of distracted driving, but also dangerous cornering, tailgating, and other near misses and close calls.
Drivers with the best scores get rewarded with gift cards and cash prizes, as well as cookouts for the branches with the best results.
“In September, I was at three different branches doing cookouts for those branches that had the three highest scores for the entire year,” says DiMascola. “We take every opportunity we can to celebrate.”
And the scores and dashboards are also used to inform the coaching sessions that managers conduct with all the drivers. “I am one man when it comes to the safety department,” he says. “We rely on branch managers and district managers to help us to deliver that safety message.”
Having an AI-powered distracted driving camera was like putting a driving coach in every truck — without having to hire 600 additional employees, he says.
To keep from being a distraction itself, the system beeps after 2.5 seconds of distraction. If the driver doesn’t address the issue, then it goes to a second level, with a series of beeps. “That’s followed by a soothing voice that says ‘distracted,’” says DiMascola. “Anything over five seconds — and that’s a heck of a time to have your eyes off the road — there’s a series of beeps that’s pretty annoying.”
Over time, the percentage of alerts that go ignored goes down, as drivers change their behavior, he says. Herr’s originally planned to deploy 100 cameras the first year, 150 the second year, and the rest in the third year. “But we quickly realized all the benefits of the cameras,” he says. “So we went back to the senior leadership to ask for the capital to be fully operational in the first year.”
Between when the program started in 2018 and November of 2021, the number of medium-to-high distracted driving events went down by 70%, he says. The number of close calls went down by 22% — and the number of collisions went down by 44%.
“We are a self-insured company,” says DiMascola. “The savings we realized almost immediately were off the charts. Prior to these cameras, we spent a lot of money on accidents that weren’t even our fault.”
For example, two weeks into the program, a professional fraudster deliberately drove into one of the company’s trucks at a four-way intersection and claimed that the Herr’s driver was at fault. The fraudster refused medical treatment at the scene and then claimed to have neck and head pain.
DiMascola pulled up the videos. “We were able to say that we had the technology, and what actually happened wasn’t even close to what they were claiming. In years past, we’re paying that claim.”
Delivery trucks are only a small part of Herr’s fleet of vehicles. There’s an equally large number of cars and vans used by field managers, merchandisers, and part-time employees. “In the not-so-distant future, we’re talking about a camera in every single vehicle that Herr operates on the road, myself included,” says DiMascola.
But without having recognized the importance of driver buy-in and Herr’s efforts to ensure drivers were central stakeholders in the initiative, those benefits may very well not have been realized.
Successful AI deployments require trust
To encourage that buy-in, experts argue that it might be better to reframe AI as “augmented intelligence.” That’s because the goal of AI, especially given the technology’s current limitations, is not to replace humans but to help them.
“Taking this approach isn’t just good for people,” says Dan Diasio, global artificial intelligence consulting leader at Ernst & Young. It’s also good for companies, he says. “First, you’re able to meet technology where it is. Technology is not perfect yet, so we can design processes with humans in the loop. Second, it sets a lower bar of expectations so you can start to put things into production faster. Third, as you design a process with humans in the loop, you are by default designing a process with trust in mind.”
Lack of trust is one of the biggest barriers to AI adoption, he says. “We found that less than 10% of proofs-of-concept make their way into a production environment. And that’s because of trust.”
Even at leading companies, trust can be an issue. According to a recent Cognizant survey, only 51 per cent of respondents at companies that are considered leaders in AI said they trusted the decisions made by AI most of the time. For companies lagging in AI adoption, only 31% of non-leaders said they trusted AI decisions.
Lack of trust can lead to a vicious cycle. Without trust, AI projects see low levels of adoption, which leads to lower business impact, which, in turn, contributes to lower levels of trust in AI.
But putting AI in the role of an assistant rather than a replacement helps build trust. Eighty percent of companies in the “AI leader” category in Cognizant’s survey saw higher potential in AI when it was used to augment human decision-making — compared to just 30% of companies in the “beginner” category.
Alliance Data looks to users to drive AI
For payment card, savings, and lending company Alliance Data, establishing that trust in AI began with a grassroots movement.
The company, which backs more than 40 million credit cards from retailers such as Ikea, Ulta, and Victoria’s Secret, saw the hype around AI, machine learning, and automation in 2018, and wanted to take advantage of it.
“We needed to find ways to work better, smarter, faster,” says Wes Hunt, the company’s chief data officer, who saw AI as an opportunity for operations center staff to spend their time on solving more complex cardholder problems instead of routine such as gathering information and manually moving data between systems.
But instead taking a top-down approach, Hunt and team asked the business units to tell them what problems they needed solving. “What came back were hundreds of ideas,” he says. “We jointly prioritized them.”
And the end users who were the most excited about the projects sat down with the data science team to design new solutions.
“That was the philosophy we used — demand-driven ideation from our partners,” he says, referring to his internal clients, leaders of other functional areas in the company. “What we found with the bottoms-up approach is the engagement level is really high. When we’re designing the way the new intelligent automations work, the guidance comes from those who are sitting side by side with those machines.”
One early project involved gathering external documentation as part of Alliance Data’s know-your-customer new account opening process. It required the use of natural language processing to read documents and computer vision to extract information from computer screens where APIs were not available.
Then, in 2020, the company began upgrading its risk models and other analytics to include more advanced machine learning algorithms to improve their accuracy and predictive value. More recently, the company has begun using AI to help improve employee performance. That has been particularly useful with the shift to working from home.
Previously, human supervisors would monitor employee performance. Some of that work has been intelligently automated. “Now employees can manage their own productivity,” Hunt says. Instead of micromanaging their direct reports, supervisors now focus on higher-level coaching, he says.
“It’s changed the coaching style that’s performed by the human supervisors,” he says. “They’re moving away from micromanagement and metric-driven management and shifting to empathy and quality.”
Today, he says, 300 brands or business processes are using machine learning or AI in production at Alliance Data.
“The secret ingredient is collaboration,” says Hunt. “The willingness to be demand-driven, to have ideas come from everyone, and to widely engage teams in the development of solutions — I think that’s what’s unique about Alliance Data. That collaboration matters.”
AI technology is hard, but it’s the human factors that make the biggest difference, experts say.
Organisations that are successful with AI projects engage with their employees, says Natalia Modjeska, research director for AI and intelligent automation at Omdia. “It’s about augmenting human capabilities, not about replacing people.”
When companies roll out AI in a heavy-handed, top-down way it can create resentment and resistance. “People fear that AI will replace them,” she says. “But if you treat people as partners on this journey rather than a target for replacement, then they behave differently.”
Yes, some jobs will become obsolete or will change as a result of AI. But this isn’t a new challenge for organisations, she adds. “We’ve faced similar tasks before in the past and hopefully by now we’ve figured out how to handle it,” Modjeska says. “Look at typists and clerks. We had to figure out how to handle it 60 years ago, and 20 years ago, and 10 years ago.”
Solutions include upskilling or reskilling employees, or phasing out jobs as people retire.
“Some of these are really hard decisions to make,” she says. “But as long as you invest in people, and treat them as partners and not as a cost centre or an enemy, these transitions go a lot smoother and benefit everyone.”