AI and HR:

Pinpointing its Potential and Learning its Limitations

Dread or desire? Avoidance or excitement?

These days, the topic of AI elicits strong opinions and emotions. There’s even a new syndrome called “AI Anxiety!” But when you leave the hype behind, smart leaders know they must stay ahead of the AI curve to remain relevant in today’s changing technological landscape.

The ramifications for HR are significant. Five years from now, HR processes may look very different. As an employer, HR manager, or benefits administrator, it’s critical to take a deep, nuanced look at both AI’s potential and its pitfalls.

The History of AI

Rudimentary AI has been around for seventy years. In 1950, Theseus, a remote-controlled mouse, could escape a labyrinth and remember its path. That was interesting but didn’t yet have real-world applications. Slowly, technology advanced into the 21st century with higher-level systems that could meet or exceed human performance.

But until five years ago, AI was only used for recognition or classification, such as voice recognition, visual recognition, or natural language processing. Google Photos, for example, can classify your pictures, and your smartphone transcribes your dictations.

The recent explosion is in generative AI, where a computer uses complex analytics to create speech or visuals. We’re already used to our email editors auto-completing our messages. Natural language generation, as in Chat GPT, is simply a more advanced version of the same principle.

An advanced AI system can now collect data, analyze problems, and provide solutions at lightning speed. Most importantly, it can present those solutions in natural, contextual language to an end user.

AI’s performance today was science fiction just one year ago. It’s here to stay, and its capabilities are improving at breakneck speed.

AI’s Business Implications

AI’s development has life-changing implications across all societies and industries. Broadly speaking, businesses can benefit from AI in three ways:

1. Scalability - AI efficiently automates repetitive tasks, manages schedules, and provides reminders. It quickly processes and analyzes large amounts of data. And it can do all that at scale without extensive hiring, onboarding, and adjustment.
2. Cost savings - Human capital is expensive. AI’s efficiency and automation can represent significant savings.
3. Personalization and engagement - Are you old enough to remember prompting your computer in DOS? Would you ever willingly do that today? A similar UX revolution is happening now. Today’s users expect mobile-friendly, on-demand, personalized apps. It’s how they shop, bank, and even seek medical advice. AI can provide those experiences.

In other words, AI allows businesses to grow quickly, cheaply, and with improved employee and client engagement. Within that big picture, each industry will apply AI differently to achieve its goals.

Applying AI to HR

Getting specific to HR, how can admins capitalize on AI’s value? The most pervasive current usage is in talent screening and selection. AI can quickly and efficiently assess resumes against job specifications to find the best-fit hires, saving time. AI proponents add that it eliminates human bias. (More on that later.)

Yet AI can go far beyond recruiting. Experts list multiple, varied uses for AI:

1. HR Process Automation: AI can automate repetitive HR tasks, such as payroll processing, leave management, and ben admin. This is especially helpful during busy times such as Open Enrollment.
2. Employee Onboarding and Training: Chatbots and virtual assistants can provide personalized onboarding experiences, answer FAQs, and deliver ongoing, on-demand training.
3. Predictive Analytics: By analyzing historical data and current trends, AI can predict employee turnover risks, future staffing needs, and resource allocation.
4. Performance Evaluation and Feedback: AI can help managers efficiently analyze large quantities of employee performance data.
5. Employee Engagement and Satisfaction Surveys: AI can use sentiment analysis to evaluate employee survey responses and identify patterns.
6. Employee Well-being and Health Monitoring: AI-powered tools can monitor employee wellness for signs of burnout or stress.
7. Exit Interviews and Turnover Analysis: AI can help HR refine its employee retention strategies by analyzing exit interview data for common themes.
8. Employee Benefits Optimization: AI can analyze employee preferences and usage patterns to customize benefits packages. It can also provide on-demand, responsive in-app benefits support.

The Elephant in the Chat Room

All these applications represent significant time and cost savings. If taken at face value, AI can improve employee productivity, satisfaction, and work-life balance, leading to higher company profits. IBM, for example, reported that its HR team saved 12,000 hours in 18 months after using AI to automate 280 tasks.

Keep in mind, however, that IBM sells AI solutions. In fact, most HR-related AI stats are from self-funded studies funded by AI companies. Still, even considering bias, there’s tremendous potential for AI in HR.

But just because a company COULD use AI, does that always mean that a company SHOULD use AI?

Maybe, but with caution. AI raises ethical and legal questions in three critical areas.

1. Legal Issues

AI providers and users are just beginning to discover its legal difficulties. Although AI touts itself as bias-free, that’s not necessarily true. An AI solution will have as much bias as the person(s) who trained it.

New York’s AI law already requires employers to have their hiring and promotion tools audited for bias by an independent third party. The Equal Employment Opportunity Commission (EEOC) warned employers that their hiring tools could be violating laws, such as Title VII and the Americans with Disabilities Act (ADA). Worse, employers will be held liable for any biases in their AI systems. Therefore, it’s crucial to train and test a bot thoroughly before implementing any new AI technology.

In line with these concerns, executives from Amazon, Alphabet, Meta, Microsoft, and OpenAI were recently called to the White House to commit to AI transparency and security. The companies agreed to both internal and external testing of their systems.

In addition, AI generative content is scraped from existing sources on the internet and is, therefore, susceptible to plagiarism. Checking for accuracy is also strongly recommended since ChatGPT has been known to make up facts when information is scarce.

2. Confidentiality Concerns

While legal issues stem from AI’s output, privacy concerns revolve around input to generative AI. For example, a project manager wants to summarize yesterday’s meeting in bullet points and create a list of action items. That task usually takes an hour or more, but it’s done in seconds with AI. The manager inputs yesterday’s meeting notes, adds some instructions, and ChatGPT writes the summary.

What most people don’t realize is that the information you put into AI is saved and may be generated as output for another user. Sensitive and protected information should never be fed to generative AI. Instead, company-specific parameters should be generalized or not used at all.

The problem is spelled out explicitly on ChatGPT’s website: “ChatGPT… improves by further training on the conversations people have with it” (J, 2023, para. 11). You can disable this function, but your inputs are still retained for 30 days (J, 2023, para. 2).

3. The Human Touch

Many employees fear, perhaps legitimately, that AI will make their jobs obsolete. There are already reports of mass employee layoffs due to AI replacements. Lower-level office jobs, such as customer service, are most at risk.

But do AI bots satisfactorily replace humans? The jury is still out. AI lacks empathy and can misinform users. Shopify’s July 2022 layoffs led to customer service delays and overwhelmed teams.

In an ideal vision of the future, AI doesn’t replace humans. Instead, it takes away time-consuming tasks to free them up for high-level work. While AI has many advantages, businesses must preserve the human touch for:

• Emotional support
• Complex decision-making
• Community building
• Physical presence
• Sensitive or confidential information.

At the same time, they can capitalize on AI’s strong points:

• Providing real-time and on-demand content
• Analyzing and summarizing large amounts of data
• Automating repetitive tasks
• Answering frequently asked questions.

Is AI coming for your job? Probably not, but with appropriate safeguards, it could become your trusty assistant in emergent and exciting ways.