Why Intelligent Finance Demands a Human-Centered Approach to AI

Erik Brynjolfsson, a professor at the Stanford Digital Economy Lab  and co-founder of Workhelix, Inc., has some advice for finance professionals concerned about whether artificial intelligence (AI) will replace their jobs: “AI is not going to replace CFOs. But CFOs who use AI will replace those who don’t.”

I had a great discussion with Professor Brynjolfsson recently at the Workday Future of Intelligent Finance Tour in San Francisco. The event—hosted in partnership with American Institute of Certified Public Accountants-Chartered Institute of Management Accountants (AICPA-CIMA) and Accenture—was designed to help finance and IT professionals leverage AI, machine learning (ML), and other advanced technologies to accelerate their journey to becoming finance futurists.

Brynjolfsson, who also serves as a senior fellow at Stanford’s Institute for Human-Centered AI, gave attendees a crash course on how AI will reshape the future of finance as we know it. Having just come from a visit to Washington, D.C., to brief the White House and the Council of Economic Advisors on the impact of AI, Brynjolfsson has a front-row seat to the potential for AI to change workforce dynamics. 

“In the early stages of AI, we knew that low-skilled labor would be impacted,” he told me. “As the technology has progressed, we’re seeing that professional workers—CFOs, accountants, lawyers, doctors—are going to be very much affected. Affected doesn’t necessarily mean automated. It doesn’t mean replaced. Rather, AI is going to augment your abilities and give you the freedom to do new things.” 

I couldn’t agree more with that positive assessment, which is why we’re investing so heavily in generative AI as the innovation engine for Workday Enterprise Management Cloud. As the group general manager for Workday’s suite of solutions for the office of the CFO, I see generative AI as an enabler of finance’s dual role as value protector and value creator.

Value protection tasks such as internal audit, risk management, and regulatory compliance can benefit from AI’s ability to comb through vast amounts of data to detect and surface anomalies before they destroy a company’s value or market reputation. Large language models (LLMs) can be trained to stay updated on the latest financial regulations, ensuring that compliance is always up to date. They can also assist in risk management by identifying patterns in data that might signify potential risks and by building robust risk models that consider a wide array of factors.

AI can also help drive value creation, whether that’s automating routine tasks to drive cost savings or enabling top-line growth. LLMs, for example, can help predict customer behavior, create more accurate forecasts, and improve scenario modeling by processing a large volume of data and considering a multitude of variables. Augmenting the capabilities of financial planning and analysis (FP&A) teams would allow them to prepare for a wider range of potential outcomes, making planning more resilient and adaptable to changing market conditions. 

Thanks to productivity gains like these, Brynjolfsson predicts that generative AI will be bigger than any of the technologies we’ve used over the last 20 years. He provided an example from new research he led on how generative AI is boosting call center productivity. 

“A couple of years ago, we teamed up with a company founded by a Stanford professor and graduate that helps call centers do a better job,” he said. “And what we found was that the operators who used the AI model were dramatically more productive and more successful than the ones who didn’t. In fact, the least experienced operators became 35% more productive.”

Brynjolfsson went on to explain that the model learned from the most successful operators, listening in on their conversations and identifying phrases or suggestions that improved customer sentiment. The model then passed on those skills to the newest operators. “That’s the kind of tacit knowledge that was previously really hard to automate.”

Generative AI and the CFO Role

Gartner analysts recently noted that CFOs are best positioned to help lead the implementation of generative AI in corporations because they have more insight into opportunities to leverage the technology to reduce costs, improve productivity, and increase revenue streams. “The CFO should be on the frontier of the AI revolution,” Brynjolfsson said. “CFOs understand how to work with unstructured and structured data and do sophisticated analyses on that data, which is why they can make such a big impact.” 

Brynjolfsson also sees human resources teams benefiting. “I did an estimate a few years ago and found that the value of human capital in the U.S. economy is a little over $200 trillion—10 times the value of the gross domestic product (GDP). But the problem with human capital is it’s very poorly measured and understood,” he said. “There are a lot of intangible skills in there. AI’s large language models can do a lot to capture and understand the value of your human capital.” 

Given AI’s power to disrupt the economy, I asked Brynjolfsson about his take on the mood in Washington around regulating AI.

“I came away from my trip really impressed with how up to speed the government officials I met with were on generative AI,” he noted. “They understand that there’s a tidal wave coming that will be bigger than the impact of the pandemic on remote work, and they are taking it very seriously.” 

Workday and Responsible AI

At Workday, we’re committed to responsible AI. We’ve been a vocal and early proponent of AI regulation that builds trust and enables innovation, and we support an ethical approach to AI that is grounded in trust and transparency. To leverage AI effectively, it’s critical to ensure that the data behind an AI recommendation is clean, accurate, and trustworthy. Given that more and more finance functions own the responsibility for data strategy at their companies, I also believe it will fall to the CFO and their team to ensure, through good data governance and management, that the data companies use to coach AI models is trustworthy.

Brynjolfsson closed out our chat on a positive note, predicting that AI could potentially double the productivity rate currently estimated by the Congressional Budget Office over the next decade. He also sees AI giving us more resources to address challenges we face on the healthcare front, such as cancer, and on the educational front, such as personalized education. The reason is AI’s ability to unlock human potential versus just seeing it as cost-saving automation. 

“Any one of you who has tried to call an automated voice response system knows it can be very frustrating, especially when there’s a long tail of questions that we ask that aren’t common,” he said. “We humans are much better at dealing with exceptions than machines, so a good partnership is where AI can answer the more common questions and humans can deal with exceptions. AI has a much higher upside in terms of creating additional value than simply trying to take costs out.” 

Read more of Workday’s insights on how AI and ML are driving the future of work.