Companies are spending millions to get AI into the hands of workers. Payback won't be quick

• PwC is rolling out a customized version of OpenAI generative AI technology to 75,000 workers, a move its chief products & technology officer Joe Atkinson says reflects the fact that "AI will change everything."
• Growing the U.S. economy at a healthy clip, even as workforce participation levels drop off, will mean that employee productivity has to increase and AI is one method to achieve this.
• CFOs worry about the high cost of the new technology, but Atkinson says the No. 1 challenge isn't the business case versus the price, it's how to prepare a workforce for its usage and at scale.  

Surasak Suwanmake | Moment | Getty Images

Generative AI tools are being rolled out to thousands of employees across companies.

PwC US, as an example, is introducing ChatPwC — its own internal generative AI tool powered by OpenAI and running in a secure Microsoft Azure environment — to each of its 75,000 workers by the end of this year.

"The hyperbole around AI is more than I have seen with anything in 30 years, but it's well-earned. It will change everything," said PwC chief products & technology officer Joe Atkinson, a member of the CNBC Technology Executive Council, who guided chief financial officers through an AI strategy session at last week's CNBC CFO Council Summit in Washington, D.C.

But will it be worth the cost? 

That's a question that chief financial officers and C-suite peers continue to debate as AI deployment becomes a priority across sectors of the economy. Many business leaders say it's a matter of when, not if.

The productivity-boosting benefits coming from artificial intelligence are well within sight, yet 44% of CFOs surveyed last week at the CFO Summit believe deploying tech tools like ChatGPT will cost them more money than it saves them over the next 12 months. The remaining 56% believe AI will save them money in the near-term.

Eric Kutcher, senior partner and CFO at McKinsey & Co., counts himself as one of the finance chiefs who thinks ROI will take longer than a year.

"We're going to run a bunch of experiments and some are going to fail," he said while speaking at the CFO Summit. "When you look at ROI over a few years, sure it's going to save us money. But if I look at it over the next 12 months, we're going to spend more than we're going to get back because we don't know how the large language models are going to evolve. I think we're going to have a number of good and noble failures."

McKinsey has created its own AI platform called Lilli (named after the first female partner at the firm) that enables consultants to do "in minutes what it would have taken them weeks to do," Kutcher said. Lilli is now being deployed among 45,000 employees, and he said "It's amazing how quickly people adopted it and how much more productive they are," noting that consultants are using it to, among other things, build client presentations.

Growing the U.S. economy at a healthy clip, even as workforce participation levels drop off, will mean that employee productivity has to increase, and finance chiefs view AI as a prime tool for making that happen.

Atkinson says the tool can help handle many of the daily administrative tasks for employees, enabling them to dedicate their valuable time to more strategic projects, allowing them to maximize their productivity and value for the clients they serve. PwC opened its internal "AI factory" a year ago and within six weeks, the firm had more than 3,000 use cases for internal uses and clients, and "very few don't have merit," he said, adding that many also came from individual contributors rather than leaders. 

Atkinson says organizations need to keep an eye on cost of licensing as well as deployment costs as the major technology providers are investing heavily in AI tools and enhancements to their existing products (including core productivity suites, finance applications, CRM tools and Human Capital applications). Over time, the large language models will become commoditized, and the true value will reside in the customization at the firm level, but it could be a year or more before that downward cost curve begins, he said. For now, the No. 1 challenge isn't the business case versus the price, it's how to prepare a workforce for its usage and at scale, he added.

When it comes to physicians becoming more productive and efficient, Anna Bryson, CFO at healthcare tech firm Doximity, said AI is a powerful tool.

"Doctors get a tsunami of new information that they have to learn all the time," she said at the CNBC CFO Summit.  "We're focused on using AI and machine learning to help make this easier for them."

At the same time, she said Doximity is deploying AI to lessen the administrative burden that doctors face. She cited the statistic that for every one hour physicians spend with a patient, they spend two hours on administrative tasks like insurance claims and chart maintenance. AI is reducing the time doctors spend on those non-patient tasks.

Similarly, bringing greater efficiencies and insights to clients through AI is a primary goal of data analytics giant Palantir Technologies, said CFO Dave Glazer. He said that his firm has been using machine learning for years, and is now bringing AI to enterprise data in order to make better operational decisions.

He acknowledged that businesses are in an AI "hype cycle" but that it's "creating a tailwind that tells me this tech is here to stay." As a result, Palantir has re-oriented its go-to-market strategy by offering current and potential customers one- to five-day AI "boot camps." The firm urges these companies to attend the boot camp with their data, "put it on our platform and build out use cases in an incredibly short amount of time."

As for where and how CFOs should get started with AI, Bryson said Doximity is taking a slow and steady approach. She said Doximity has already invested in AI that helps with sales forecasting. "Even if we decide over the next year this software isn't that helpful to us, it will be worth it because not only will we have a tool that can assess our pipeline health, but it can also help us with market trends by account and by industry," she said. "That's worth experimenting with."

And while McKinsey's Kutcher believes the costs associated with AI will eventually come down, he describes it as a "multi-year journey, not something that is happening over the next six months."

Still, he added, "I don't think this is about the cost efficiency of the technology in the short run. I think this is about what you do with the technology and how do you recognize it in either productivity or revenue."