Not yet a Member?
HR Magazine is highlighting the next generation of HR leaders.
Is your employee handbook ready for the New Year? With SHRM’s Employee Handbook Builder get peace of mind that your handbook is up-to-date.
30+ HR education programs, including 4 NEW programs on hot topics, are available for registration.
Join us in Chicago for the latest trends and technology in talent management, and what to expect in the future.
Artificial intelligence (AI) is transforming the workplace—eliminating some jobs while creating new ones. HR professionals need to both familiarize themselves with the emerging technology, be ready to educate employees about it and yet also question its application when it threatens to reinforce deeply entrenched biases.
"Don't be intimidated and [do] ask questions," said Margaret Keane, an attorney with DLA Piper in San Francisco. "One mark of truly understanding a subject is being able to explain it coherently to others. Processes are changing, and you need to understand them."
What Is AI?
AI means different things to different people. Some balk at the term "artificial intelligence," preferring instead "augmented intelligence." Artificial intelligence remains the more common terminology.
For Keane, the definition that makes the most sense is from the MIT Technology Review: "an evolving constellation of technologies that enable computers to simulate elements of human thinking, including learning and reasoning."
That may include machine learning and deep learning.
Machine learning "is a type of AI that provides computers with the ability to learn without being explicitly programmed," she said, referring to WhatIs.com's definition. "Machine learning focuses on the development of computer programs that can teach themselves to grow and change when exposed to new data."
And Keane said that, according to graphics technology company Nvidia, "deep learning is the fastest growing field in machine learning. It uses many-layered deep neural networks to learn levels of representation and abstraction that makes sense of data such as images, sound and text." An example of deep learning is an application improving a car's ability to detect objects and recognize sound.
Big data might be harnessed by machine learning but doesn't have to be, said Zev Eigen, Ph.D., an attorney with Littler in Los Angeles and its global director of data analytics. A "neural network model" replicates the human brain and can compare numerous fields, such as where someone went to school, their years of relevant experience, and civic activities, and arrive at the best output, predicting what person would be a good hire, how long the individual is likely to stay and if he or she is amenable to change.
But if the input includes a protected category, like race, gender or age, machine learning may unlawfully discriminate. Even if protected categories aren't in the input, some computing through machine learning may result in a disparate impact, so stay on your toes, he cautioned.
Though an employer does not intend to discriminate, it can face "huge liabilities" if HR decisions based on seemingly neutral factors have a statistically significant effect on protected groups, explained Corey Goerdt, an attorney with Fisher Phillips in Atlanta.
The potential for bias is a big workplace compliance concern with AI. "Employers can't assume that candidate pools generated through AI tools are bias-free, as AI tools may have 'baked-in' bias," Keane noted. But, on the other hand, "AI may be a tool to reduce bias."
"Companies like Twitter, Microsoft and Hilton Hotels have started using technology to remove the human bias inherent in subjective hiring decisions," Goerdt said. "For instance, technology developed by HireVue allows companies to make hiring decisions by analyzing speech patterns and physical gestures in standardized video interviews. HireVue also uses its expansive database of video interviews to evaluate hiring decisions based on subsequent performance of hired candidates." But what if the speech patterns and physical gestures of one protected category of individuals differs from others?
Validation of the selection tools and procedures for selecting candidates and monitoring for adverse impact is key to addressing such concerns, Keane noted.
"Machine learning databases are often cloud-based, which gives the machines access to large sources of information being sourced from different people at different locations," said Mike Nelson, an attorney at Sutherland Asbill & Brennan in New York City. "This 'hive mind' approach gives AI the ability to make the best decision not strictly based on what a single machine has learned, which may be quite limited, but on an entire database of scenarios, experiences and solutions compiled by however many machines are operational and connected to the cloud. The greater and more widespread the active information source, the more likely the machines are to reduce biases that may pervade a particular workplace."
"Regardless of the tool used, selection criteria need to be job-related, meaning that they measure a trait or skill that is both job-related and consistent with business necessity," Keane said. "Algorithms can get so complicated that people may not know what they are measuring. The selection criteria need to be valid, meaning that there is a causal relationship with the ability to perform the job and not simply a correlation with common traits shared by employees in this role."
For example, correlation is the observation that people who grew up in certain suburbs do well in a job; causation is the finding that people who have studied statistical analysis and data visualization do well in a data analytics group.
In addition, employers should be sensitive to privacy concerns. AI may go into systems that include personal information and review e-mail, scraping key indicators to see how much nonwork e-mail employees have and how much time they're spending on it, Eigen said. Over time, employers may need greater transparency to reduce concerns, or lack of trust may instigate the need for regulation, he noted.
Oxford researchers predict nearly half of U.S. workers will see their jobs automated over the next 20 years, reports the MIT Technology Review.
"AI will change the workforce in ways that are predictable and in others that we can't foresee," Keane said. "The certainty is that jobs will be lost, other jobs calling for different skillsets will be created and roles now performed by knowledge workers will be different as AI is used to augment human skills."
The key word is augment. Eigen said decision-makers like HR won't be replaced by AI, but will use it to enhance their effectiveness.
AI isn't just something in the distant future. Already it's changing work, whether it's chatbots answering questions about jobs, technology that solicits resumes and matches people to jobs with people overseeing it, or robotics that sometimes are powered by AI, Keane said.
[SHRM members-only toolkit: Managing Organizational Change]
The best examples of how AI already has changed the workforce are in the hiring process with new sourcing, screening and ranking tools, Keane observed.
In the legal sphere, AI already is running search and predictive coding—new tools to perform due diligence tasks, she noted.
The advent of AI has powered personal assistants for scheduling and administrative tasks that will lead to changes in administrative roles "that have already seen radical change from advances like voice mail and personal computers," Keane said.
As for the next generations of workers, AI is likely to result not only in the disappearance of some jobs common today, but the emergence of new ones. The World Economic Forum's The Future of Jobs report estimates that "some 65 percent of children entering primary school today will ultimately end up working in completely new job types that don't yet exist."
Was this article useful? SHRM offers thousands of tools, templates and other exclusive member benefits, including compliance updates, sample policies, HR expert advice, education discounts, a growing online member community and much more. Join/Renew Now and let SHRM help you work smarter.
You have successfully saved this page as a bookmark.
Please confirm that you want to proceed with deleting bookmark.
You have successfully removed bookmark.
Please log in as a SHRM member before saving bookmarks.
Your session has expired. Please log in again before saving bookmarks.
Please purchase a SHRM membership before saving bookmarks.
An error has occurred
Recommended for you
CA Resources at Your Fingertips
SHRM’s HR Vendor Directory contains over 3,200 companies