As artificial intelligence dominated conversations at this year’s HR Tech conference, new research shows a significant gap between AI experimentation and formal implementation within HR processes.

According to the 28th edition of the HR Tech HR Systems Survey from Sapient Insights Group, 31% of organizations now use AI within HR processes, up from 24% the previous year. However, a more striking finding reveals that 80% of HR professionals use AI tools personally for work tasks, yet only 14% pay for these tools themselves.

“There’s a difference between whether I’ve embedded it and I’ve built a process around it and other people in the organization are accessing it versus I’m using my own version of AI,” said Stacey Harris, chief research officer and managing partner at Sapient Insights Group, in an interview with HR Executive HR Tech Editor Jill Barth at HR Tech.

Redefining AI

The survey data shows an interesting shift in how organizations define AI. Harris noted that companies are moving away from counting machine learning and predictive analytics as AI adoption, now requiring generative or agentic capabilities to meet their definition.

She said that organizational teams, rather than AI vendors or tech businesses, are the ones reshaping the definition of what counts as AI. Over the past year, the framing has shifted noticeably.

The kinds of machine learning that major corporations relied on, including predictive analytics, used to be described as the core of artificial intelligence. That view is no longer dominant.

Many companies frequently claimed they were actively using these AI technologies. Harris observed that while this adoption was common, it didn’t necessarily reflect the full potential of AI.

However, it seems that the perception has shifted recently. Now, the prevailing sense among these organizations is that true artificial intelligence must involve something more advanced. Specifically, this indicates some form of generative and agent-like intelligence that can act autonomously and create new content or solutions, rather than just analyzing existing data and making predictions.

Learn more about the study: Sapient Insights Group’s 28th HR Systems Survey: Top solutions, trends and what’s next

Cost becomes the primary barrier

For enterprise organizations with more than 5,000 employees, cost has emerged as the top barrier to AI implementation. This category jumped from 22% citing it as their primary concern last year to 44% this year. “They cannot afford what’s in front of them,” Harris said.

The survey found strong resistance to extra AI fees, with 57% of respondents uninterested in paying more for AI capabilities. Around 40% said they might consider additional costs if there were a clear and compelling ROI. Only 5% were outright willing to pay extra.

HR’s strategic rise

The survey revealed a significant jump in how CEOs and board members view HR’s strategic importance, rising from 34% to 54% year over year. Harris attributes this to IT departments gaining newfound respect for HR’s expertise in managing unpredictable resources.

“They’ve really gone through a whole year of learning what it’s like to manage a resource that doesn’t always do what you think it’s going to do,” Harris said. “HR has been doing this for a long time.”

Read more: What are the key findings of the 2025 HR Systems Survey?

Business priorities driving HR tech decisions

Two major forces are shaping business decisions: Forty-one percent of organizations prioritize revenue growth, while 70% focus on optimizing efficiency and reducing costs. These priorities are bringing HR and finance into closer collaboration, though not always smoothly.

“HR is getting stuck in the middle a little bit with budgets and who owns what, whether the data is valuable for AI or is it valuable for business outcomes,” Harris said.

When organizations do implement AI in HR, the value proposition centers primarily on efficiency gains and faster information access for managers and executives. Examples include extracting information from documentation and databases that were previously difficult to access, particularly in benefits administration.

Harris noted that the most transformative applications of generative AI are still largely unrealized. She emphasized that its true potential lies in assisting people, helping them collaborate more effectively and think in new ways. “And I don’t think we’re there yet,” she added.