Increasingly, HR departments are staffing data science functions to work with standard metrics, predictive tools and other intelligent applications with the overall responsibility for all of HR’s data.
But in a session at the HR Tech Conference titled “Intelligent HR Tools: An Introduction to AI for HR,” John Sumser, founder and principal Analyst at HRExaminer and HRE columnist, says technology can’t solve every HR problem.
“Bias is not a problem you can solve with a machine,” Sumser said. “It’s built into humans and they use it to speed through decisions. You can do some interesting things, but technology will always contain the bias of its designers.”
Currently, he said, some of the basic problems with using AI in HR include: fidelity, or a correspondence between the model and reality; and latency, or a time lag between reality and data.
So where to begin?
“Resist the temptation to begin without a clear vision,” Sumser said, adding that companies should also inventory their available data, do industry research, and start an ethics team and a governance team in-house.
While the pace of technology will continue to increase and eliminate some of these problems, Sumser said, it’s still important for leaders to realize their own limitations.
“You don’t get ahead by knowing everything,” Sumser said. “You get ahead by saying ‘I don’t know.’ The sooner you can say that, the sooner you can get your answer.”