Payroll Has a Friend in Generative AI

Generative AI has many use cases involving payroll as a tool to help with compliance and respond to industry trends, a consultant said May 8.

Kathy Soderstrom, a managing director at EY, highlighted the increase in the complexity and amount of compliance needs and privacy laws, scarcity of payroll talent, and payroll’s increasing role as an advisor to company executives as payroll trends behind the increasing use of AI.

She also mentioned general labor trends as including inflation, job openings and retirements, and expanding technology in her comments at the 42nd Payroll Congress in Nashville.

Her descriptions of how the payroll industry might involve next included integration of technology and automation of processes such as source-to-gross calculations and identifying and correcting other errors.

Soderstrom defined “generative AI” as AI that can create new content based on the data it has been trained on. She said that it help with analyzing content, making decisions, and increasing productivity in other ways.

For example, many employers have a large number of wage codes that they have accumulated over the years, and when changing payroll platforms AI could interpret wage types and suggest a category match, but with a human making the final decision, Soderstrom said.

The technology could also analyze global payroll data for “missing and unusual” payments to cross-border employees and to detect errors, Soderstrom said. She mentioned a client story where an employee moved countries after annual bonuses were determined but before they were paid, causing a currency change that threw off the amount of the bonus compared to what was determined.

AI chatbots could also take over answering employee questions, especially common ones such as when Forms W-2 are issued or about employee payslips, freeing up payroll staff for other tasks, Soderstrom said. “It’s actually labor-intensive to answer every single employee question,” she said.

Soderstrom added that a chatbot could work up to answering more complex questions such as “can you tell me why my paycheck has different withholding this cycle compared to last pay cycle?” and could also use public, company-specific, or employee-specific information to answer questions.

AI could also provide reports that Soderstrom phrased as “user-requested,” not “user-built,” because they would be generated by asking AI a question. She said “a lot of people went into payroll not planning on being technology developers,” yet often have to work out how to put together a report in payroll software.

The future of the payroll profession will see the ability to use AI and other technology become more desirable, Soderstrom said, adding that she saw an increasing emphasis on payroll using “meaningful” insights and analytics to advise the larger company.

The ability to stay on top of global payroll laws will also be important, Soderstrom said. Payroll departments should either have access to people that are proficient at compliance or become proficient themselves, she said.

“Befriend AI,” Soderstrom said. “I think we can all explore and learn from each other on how it can help our functions.”