Why 88% of Firms 'Embrace' AI But Only 5% of Adopters See Measurable Impact

That gap between 88% of corporate executives claiming that their firms 'embrace' AI while only 5% admit to achieving a measurable bottom line impact is...large.

What accounts for it, once one wades through the dense undergrowth of institutional jargon - optimizing outcomes, transforming impacts, etc - appears to be two primary factors. The broader one, which is actually positive, is that the current generation of leaders and those moving into leadership positions are the first true digital natives to achieve that level of power. These are not just 'kids' who played with desktops, they are people who literally grew up with cell phones in hand. Tech is not second-nature, it is their core lived experience, which makes adoption of a potentially epochal technology more interesting than fearsome and their ability to apply it intelligently more likely. The second is FOMO, fear of missing out. They can't say that they really know where this is headed or that they can confirm how successful it will be - and they acknowledge the hype and the risks. But they don't believe they can afford to wait to see. JL

Steven Rosenbush reports in the Wall Street Journal:

88% of enterprises declare they are undertaking an AI transformation, (but) only 5% of the AI adopters said they have succeeded in transforming a single business domain to achieve a measurable bottom-line impact. What is different with this tech era? 1) AI has proven easier than previous technologies because there are lower barriers to adoption and it lends itself to an accessible user experience for businesses and customers. 2) Companies believe they can measure the effect of AI. 3) "There is a pervasive fear at the senior leadership level that AI will fundamentally disrupt the business model." Executives acknowledge the risks and limits of AI but just see them as management challenges. The existential danger now lies in the failure to adapt.

After largely failing to grasp the internet, social media, smartphones and e-commerce quickly enough to avoid significant pain, many businesses have already gotten onboard with AI. 

Many potentially transformative applications of artificial intelligence remain hypothetical, of course. Researchers are still assessing the strengths and significant limits of reasoning models. Cybersecurity professionals emphasize the risk of connecting AI with private company data. And regulators, lawmakers and advocates want rules to protect humans on the whole from out-of-control AI.

Meanwhile, 88% of enterprises declare they are undertaking an AI transformation, according to research from McKinsey. 

I wanted to know what was different with this particular tech era, so I asked Athina Kanioura, chief strategy and transformation officer at PepsiCo, which has been deploying AI for years. 

“For the first time with a new technology, there is a pervasive true fear at the senior leadership level that AI will fundamentally disrupt the business model at its core,” she said. “The potential for gen AI to completely rewire how value is created is so profound that no large company can afford to wait.” 

PepsiCo, a food and drinks giant with nearly $92 billion in revenue last year, began by applying machine learning to areas such as demand planning and predictive asset maintenance in manufacturing facilities, then moved on to extracting insights from data collected, for example, in warehouses and manufacturing plants. The company connected autonomous vehicles in those facilities with an “AI control tower” that made work for human employees safer and more efficient.

It added generative AI to the mix about two and a half years ago in areas including research and development and innovation, logistics and taking orders from other businesses.

Athina Kanioura, chief strategy and transformation officer at PepsiCo, says every AI initiative at the company needs a clear, measurable top-line and bottom-line impact. Photo: PepsiCo

PepsiCo is also using agents from Salesforce to manage key operations including some sales functions and customer support procurement, sourcing, compensation advice and IT support tickets. “There is no reason for a human to [have] to do those type of tasks,” Kanioura said. Agents are better suited for many of the duties involved because they can more readily access knowledge from systems across the company. The human role then shifts to high-value advisory work, such as having conversations with employees about career paths or significant life events, she said.

 

The company hasn’t fired people because of the shift to AI, but AI does allow it to grow without adding staff in the ways it needed to in the past, she said.

AI has proven easier to adopt than previous technological shifts because there are lower barriers to adoption and it lends itself to an accessible user experience for both businesses and their customers, Kanioura said. While companies do need to invest in data infrastructure, governance and security, they don’t have to create a new physical infrastructure as they did with the rise of e-commerce.

Companies also believe that they can measure the effect of AI on their business and that of their customers. Kanioura says PepsiCo operates under a strict mandate: “Every AI initiative must have a clear top-line and bottom-line impact that is tracked and measured before activation.”

She declined to share specific numbers, but said the bottom-line impact on customer service, for example, is “very significant” due to increased efficiency and effectiveness. Improved service levels directly translate to higher sales to distributors and other customers, she said.

The benefit to employee experience is measured in saved hours that boost productivity.

PepsiCo’s agent deployment based on Salesforce tech is under way in the Americas, she said, with Europe, Asia, and Africa to follow, “demonstrating a methodical, global deployment strategy built on a solid data foundation and clear process mapping.”

 

Kanioura said large language models, while powerful, still lack sustained context, consistency and true agency. Considering the time, cost and talent required to solve those problems, she said PepsiCo decided to “fill those gaps” by building within the Salesforce platform, tapping into existing resources such as a semantic layer of software and application programming interfaces.

There is another motivation for companies’ speedy adoption that lurks not so far behind all the others: Executives just don’t want to wake up behind the curve again. If their sectors are going to be as disrupted as everybody is telling them, this time they want to at least have a hand in events.

Whether 88% of businesses have the experience, tech savvy and imagination to make much of AI in the near term is an open question. Some would say certainly not. 

And only 5% of the AI adopters in McKinsey’s research said they have succeeded in transforming even a single business domain to achieve a measurable bottom-line impact. But businesses remain comfortable embracing AI, according to Alexander Sukharevsky, McKinsey senior partner and global leader of QuantumBlack, McKinsey’s AI arm.

Executives acknowledge the risks and limits of AI. They just see them as management challenges. The existential danger, to them, now lies in the failure to adapt.