AI Adoption Surges During COVID-19, KPMG Finds. So Do Ethical Concerns
Real-world AI deployments surged over the past year as companies sought to remain competitive during the coronavirus pandemic, according to a new study released today by KPMG. However, even as they expanded and accelerated their AI projects, organizations expressed concerns about ethics and bias, and suggested AI might be getting ahead of regulations.
KPMG’s study, called “Thriving in an AI World,” replicates a study conducted before COVID-19 upended our world a year ago. That provided KPMG Principal of AI, Traci Gusher, a convenient baseline to test how AI deployments have been impacted by COVID-19.
“Over half the business leaders that we talked to said that AI is at least moderately to fully functional in their organization, which is a significant increase,” Gusher says. “Year over year, what we’re seeing is not only has AI adoption continued, but it actually significantly sped up, even during a global pandemic.”
On an industry basis, the percentage of organizations saying their AI is at least moderate functional increased substantially. It increased by 37 percentage points in financial services, going from 47% claiming moderate or full AI adoption in 2020 to 84% in 2021. The corresponding increases in retail was 29 percentage points (from 52% to 81%), and 20 percentage points in the tech sector (from to 63% to 83%).
AI adoption didn’t increase in 2020 and early 20201 in spite of COVID-19, according to Gusher; it increased directly as a result of it. “Over night, everyone went remote, which just in general sped up everything digital,” Gusher says. “There’s this overall digital wave that is continuing to flood us, and AI is a piece of that.”
Companies that had already deployed AI in their organizations were operating at a higher level when COVID-19 dramatically disrupted so many facets of our lives, and their competitors took notice, Gusher says. For example, companies in the consumer goods supply chain that could “scenario-plan black swan events” had an advantage over those that couldn’t, she says.
“If you had virtual agent conversational capabilities that enable you to rapidly scale your customer channels without relying on humans who had to leave the workforce, were getting sick, or having family members getting sick–if you had the digital capability to scale your customer service channel using conversational agents, then you fared better with your customers,” she tells Datanami.
“I think in general, organizations realized that AI adoption isn’t something that’s nice to have or is something that you can just play with in functional pockets,” she continues. “AI at scale is something that’s a real competitive differentiator.”
The technology sector led their peers in terms of the sophistication of AI deployments, with 52% of firms saying they are “fully functional at scale” and 31% with “moderately functional” deployments. Retail was just slightly behind, followed by industrial manufacturing, healthcare, financial services, life sciences, and government.
Technological improvements will likely accelerate AI adoption even more in the years to come in the consumer goods supply chain, Gusher says. “There are some really interesting things going on right now at the junction of 5G, IoT, and AI for warehouse and distribution,” she says. “There’s a bit of a spike going on with the use of digital twins and simulation as an area to help plan for disruption in manufacturing, in supply chain, and distribution.”
Fraud detection is a common use of AI in financial services, and 93% business leaders in that industry expressed moderate to high confidence that AI can successfully detect fraud, an 8 percentage point increase from last year, according to KPMG’s survey.
Similarly, in government organizations, 79% of leaders say they are moderately to highly confident in AI’s ability to improve bureaucratic efficiency. Comparable numbers are seen in the life sciences and healthcare sectors with regard to AI’s capability to help track the spread of COVID-19 cases, help with vaccine development, and help with vaccine distribution.
However, the survey also detected a change in sentiment in regards to the ethical concerns of AI deployments. While business leaders recognize the need to expand their AI rollouts for competitive reasons, they simultaneously seem to be uneasy with it.
“In the study we did a year ago, one of the findings we had was organizations were actually not fearing government regulation and oversight, but they were actually asking for it,” Gusher says. “When we did this year’s survey, we thought this is going to be a really interesting one to watch to see if they felt the same, if they changed at all, etc. Overwhelmingly, across every industry we talk to, the leaders want more government involvement in AI regulation.”
The 2021 study found that large numbers of business leaders think AI is moving too fast, ranging from 55% of leaders in industrial manufacturing expressing that sentiment to 35% in healthcare. Compared to the 2020 study, the biggest deltas emerged in financial services, retail, and technology, which saw 27 percentage points, 24 percentage points, and 17 percentage point increases in business leaders asking for government regulation of AI, respectively.
That sets up an interesting dynamic, whereby executives recognize the business imperative of adopting AI and want to move even faster with AI, but simultaneously recognize some of the dangers of moving too fast, Gusher says.
“We saw no evidence in the survey last year or this year that anybody wants to slow down AI adoption. But we did find people saying, hey we might be moving a little too fast,” she says. “[They] don’t want the government to stop [businesses] from doing it. I think organizations want to understand what the guidelines are so they do it properly. I think they want to understand what should the guideline be for mitigating bias? What should the guideline be for setting an ethical process on how they use AI? What should the guideline be on the types of testing and validation that has to be done in order to use AI in certain domains?”
Gusher says Congress is becoming aware of the concern about AI, and today’s KPMG study will only increase that awareness. She predicts there will be a federal AI regulation during the Biden administration. Ideally, it would come after a national data privacy law, she says.
KPMG’s survey involved 950 decision-makers who had “at least a moderate amount of AI knowledge” and at companies with over $1 billion in revenue. It was conducted online between January 3 and January 16. It carries a margin of error of 3.2%.
Editor’s note: This article was corrected. The previous year-over-year comparisons for AI adoption in several industries was incorrect. The story now reflects the increase in accurate percentage-point terms, rather than a simple percent increase, which was inaccurate. Datanami regrets the error.