This column was originally published on Jeff Neal’s blog, ChiefHRO.com, and was republished here with permission from the author.

In my last post, I wrote about the idea that artificial intelligence may begin to replace knowledge workers in the federal government. This post addresses some of the ways AI may replace HR in the not-too-distant future. The ideas I am covering are not fantasies or science fiction; they are based on actual technology that is either in use or far along in the development process.

So how might your neighborhood HR specialist be replaced by AI? And how likely is it to happen? Let’s start with the simpler processes such as proving routine information to employees and job applicants. Many of the calls HR receives are requests for information. Employees and applicants want to know the status of something, or how to do something. This type of call rarely requires in-depth analysis — it just requires someone who understands the question and knows where to find the answer.

Chatbots can handle much of this type of work, and their capabilities are advancing rapidly. Here is an example of a chatbot from Amazon Web Services. Services such as these do not rely on a user picking from a list of options, then from another list, then another, until the user eventually gets to an answer or to someone who can help. They rely on natural language, so the caller can ask the question in the same manner she or he would with a human HR Specialist.

Centralizing benefits tasks

Another part of HR that lends itself to AI is benefits. The largest numbers of interactions between employees and HR are for benefits. Workers want to sign up for health insurance, change to a different insurance plan, ask for retirement estimates, file an injury compensation claim, or perform some other benefits related task. In much of the private sector and in some agencies, the bulk of human interaction of benefits work is performed in call centers. In the past 20 years, much of the rest of it has moved to employee self-service. When we combine emerging natural language generation capabilities of AI with the trend toward employee self-service and centralization of benefits work, it is easy to see how the benefits counselor of the future will have silicon rather than flesh.

Another aspect of HR is compliance. The traditional approach to compliance is laborious audits that consume resources and offer little return when the agency is following the rules. AI will eliminate much of that work by analyzing data for transactions that do not comply with established norms. That type of analysis will also help agencies to identify issues with policies, decision-making, and other aspects of agency people programs.

Over reliance on position-filling technology

Much of what is left in HR revolves around establishing and filling jobs. The fields of staffing and position classification have been the focus of much of the process automation in HR for many years. What has happened in those areas is not artificial intelligence. In fact, I would argue that it has been closer to artificial stupidity. The technology we have used to classify jobs and fill them did not relieve HR specialists of mundane tasks and allow them to focus on providing great customer service. What actually happened was an over reliance on technology that was not intelligent, combined with poor training for HR specialists, that resulted in barely functional HR processes. Let’s take a look at each of those and see what happened, and where AI could help solve the problems.

Position classification is one of the processes that has been subject to automation for about 30 years. Because the classification process was cumbersome, required significant expertise, and often resulted in managers and employees not getting the grade levels they wanted, there was a move to automate the process. The products that were produced did not automate anything. They dumbed down the process so position classification devolved into a process of selecting the right set of words from existing job descriptions and classification standards to match a grade and job series a manager wanted.

To make the process easier, HR moved to more and more generic job descriptions (more on those later). The process gave the illusion of savings, but did not really save anything. We may have eliminated some position classification specialists, but every dollar we saved by getting rid of them was spent many times over in over-graded jobs, hiring processes that were broken because they were based upon generic job descriptions that did not produce the right talent, and other similar problems.

Hiring is another area that was automated poorly. What we see in most systems in use in the government, such as USAStaffing, is not artificial intelligence. We replaced cumbersome Knowledge, Skill and Ability essays with poorly designed applicant questionnaires that do not identify the good applicants, even if we can get them to apply. HR Specialists use canned questionnaires over and over, including garbage filler questions that make it far less likely that the best applicants will show up on a referral listing. This is one of the areas where AI has the opportunity to truly transform the process.

Many years ago, the Defense Department, NASA and some other agencies used a system called Resumix. The idea behind Resumix was great — the system would scan résumés for the knowledge, skills and abilities that were necessary for the vacant job. The problem was that the technology did not exist at the time for that to work. The result was that HR, after years of telling people they did not simply look for keywords in applications, deployed a system that looked for keywords. Resumix did not find a wide market and eventually was discontinued.

More Commentary

How 21st-century data management can help leaders provide more effective correctional healthcare services Commentary

Breaking down barriers: The challenges of federal micro-purchases for small businesses Commentary

The push to upskill the technology workforce in federal agencies Commentary

Rise of AI hiring tools likely

We are now arriving at a level of AI that can make the idea of the early products a reality. Within a few years, we are likely to see a rapid growth in use of AI to read résumés and do the screening that needs to be done. Such a move would replace the self-assessment questionnaires that have proven to be so unreliable. I believe the transformation will go far beyond HR replacing one tool with another. Instead, I expect to see these tools made available directly to hiring managers. When so much of the process can be automated, why do we need an HR specialist in the middle of the process? Why not allow managers to write and classify job descriptions, then fill the jobs? Moving beyond simple screening of the application, we should expect to see other parts of the process, such as interviewing, on-boarding and initial training, automated as well.

If that happens, what is left of HR? Is it just the helpful robot, with only a handful of people left to advise on matters such as labor relations and disciplinary actions? I think the answer is no, it can still be far more than that. HR professionals have said for years that they want to be trusted advisers. At the same time, they undermined that objective by providing substandard services that made them anything but trusted. There is a role for HR in hiring, but it is probably going to be filled by experienced recruiters who can help hiring managers find the talent they need, rather than the stop sign waving bureaucrat who says no to everything and runs a process that cannot separate the good applicants from the bad.

There is probably a role for someone who understands how organizations work and how they can be structured to get results. There is a role for the training professional who understands what kind of skills workers need, how adults learn, and how an organization can ensure that its people maintain the skills they need to succeed.

I believe the AI-enabled HR office can finally achieve the goal of being a trusted and needed part of every agency. The biggest risk is not that the technology will not work, it is that fear of risks will cause agencies to take timid baby steps, or that the defenders of the status quo will fight tooth and nail to preserve what we have now. My next post on AI will address the steps I believe we need to take to ensure that the future wins and the naysayers do not.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.