Exploring AI in benefits

You’re reading and hearing about AI everywhere—and you’re likely using it, too. AI adoption among HR professionals increased to 72% in 2025, up from 58% in 2024.

One area where it’s seeing growing use? Benefits administration. According to a 2024 survey from Mercer, 40% of HR leaders were already using AI in benefits administration, and that was expected to increase by another 40% over the next year alone.

But adoption and optimization are two very different things. To make the most of AI tools, you need to understand where and when they’re best used (and what the potential tradeoffs are). This guide is here to help you make sense of the pros and cons, so you can make smarter decisions about where AI fits into your benefits strategy.

Leveraging AI in benefits administration

Where does it fit in? What can it do? And just as importantly, what shouldn’t it do?

Here’s the gist: AI in benefits administration means using software that can automate routine tasks, analyze large datasets, and deliver personalized guidance to employees.

For example, here are a few specific areas where you’ll see AI come into play:

• Member navigation and enrollment: Using AI-powered chatbots and virtual assistants to answer your employees’ questions, walk them through plan options, and help them make enrollment decisions.
• Claims and prior authorization processing: Conducting an automated review of claims and coverage requests, which can speed up a process that’s otherwise eye-glazingly monotonous and slow.
• Population health analytics: Analyzing claims data across your entire workforce to identify trends, flag high-risk employees, or highlight gaps in care utilization.
• Benefits utilization: Identifying which benefits your employees aren’t using so you can tailor your strategy accordingly.

Those are common benefits administration tasks—and they take up a lot of time and effort when they’re done manually. For that reason, benefits administration (vs. performance management or training and development, for example) is the HR area where employers have the largest appetite for AI, with 39% of employers saying it’s a task better handled by technology than by people.

As far as what tools are used to execute the above tasks, AI runs the gamut from simple rule-based automation to sophisticated machine learning models. And because AI itself is so varied, so are its capabilities and risks.

Where AI is genuinely delivering

Let’s start with the upsides. In a recent survey, 73% of benefits leaders say they believe AI will have a positive influence on benefits administration. Here’s a look at a few of the areas where technology can make a real difference.

Cutting through enrollment confusion

Benefits enrollment is notoriously tedious, overwhelming, and confusing. Many employees just default to their choices from the previous year, rather than wading through paperwork to evaluate their options.

AI tools (like chatbots or assistants) can answer employee questions in real-time and offer personalized guidance as they compare plans and make their enrollment decisions. And, unlike human staff, AI doesn’t keep office hours—which matters, since 30-40% of employee questions about benefits come outside of normal working hours.

Additionally, workers likely feel more comfortable navigating their choices with a bot over someone from HR. Privacy and self-consciousness play a real role here. Seven in 10 employees have avoided asking HR questions for fear of looking uninformed, and more than two-thirds have kept their lips zipped due to privacy concerns. So, it’s not surprising that 73% of employees are already using AI for guidance on more sensitive topics like personal health, finance, and wellness—or that 58% say they feel safer opening up to a chatbot about mental health concerns than they would to HR.

Put simply, AI gives workers an easy, accessible option to improve their benefits literacy—and actually use more of what you (and they) are paying for.

Freeing up HR teams for higher-value work

The average worker loses up to 51% of their work hours to repetitive, low-value tasks like data entry, copying and pasting information, or managing emails. Unfortunately, this kind of routine administrative work is particularly common in HR, where teams are dealing with enrollment changes, predictable questions, and dependent verification.

Those things are important, but they shouldn’t monopolize HR’s time. AI can handle the recurring administrative tasks so benefits teams can focus on the bigger stuff—like strategy, vendor relationship management, and supporting employees.

Leveraging AI can have a real impact. For example, IBM introduced an HR AI agent (called “AskHR”) to automate routine work across benefits, payroll, and job-related questions. Productivity improved by as much as 75% in some areas.

So, it’s no surprise that leaders are paying attention. 81% of HR leaders have already explored or implemented AI to improve the efficiency of various processes within their organizations. Benefits administration, with its heavy mix of repetitive and time-sensitive tasks, is a natural place to start.

Catching problems before they become expensive ones

Some of the costliest health problems start as small, easy-to-miss clues buried in your data. An employee who hasn’t filled a prescription for a chronic condition. A sudden spike in ER visits. Employees who haven’t scheduled the preventative screenings they’re eligible for.

AI can analyze patterns like these in your claims and utilization data that would be nearly impossible for a human reviewer to flag in a timely manner (particularly across a large workforce).

This helps you catch manageable conditions early—and that’s almost always cheaper than treating a serious one later. In a UCLA Health study, proactive care management targeting AI-identified at-risk patients produced a 27% reduction in potentially preventable hospital admissions.

In this way, AI can be a strategic tool, not just a time saver. It actively improves the health of your workforce (and your costs as the employer).

Where AI in benefits gets risky and complicated

There’s no denying that AI shows real promise for benefits administration—but only when it’s deployed thoughtfully. Here are a few areas where the risks are real and worth understanding before you roll out any new tools.

Speeding up the wrong decisions

Prior authorization—the process insurers use to approve or deny coverage for certain treatments, procedures, and medications—is one of the fastest-growing applications of AI in benefits. It’s historically slow, labor-intensive, and a big administrative burden for everyone involved. That’s a problem AI seems well-suited to fix.

But faster isn’t always better. To be fair, prior authorizations were already flawed. But according to a 2024 Senate committee report, AI tools used in prior authorization have been accused of producing care denial rates as much as 16 times higher than typical. That’s largely because insurers are using these systems to issue batch denials with little to no human review.

Providers are sounding the alarm, too. In a 2024 survey, 61% of physicians reported concern that AI use by health plans is increasing prior authorization denials.

As an employer, you’re not the one handling prior authorizations. But this still matters for you. If AI steers employees away from necessary care (even inadvertently), you’re not reducing your costs. You’re just contributing to bigger claims in the future.

Widening care gaps

AI recommendations are only as good as the data they’re trained on. Unfortunately, in healthcare, that data comes with decades of demographic inequity—unequal health access, underrepresentation in clinical research, and systemic disparities in how different populations have been treated and documented.

When AI systems are fed data that underrepresents certain populations, those groups might receive lower-quality triage or care recommendations as a result. That has real consequences. A 2023 report found that minority ethnic people, women, and people from deprived communities face a risk of poorer health outcomes due to biases in medical tools and devices.

Workers don’t just need healthcare guidance—they need good and trustworthy healthcare guidance. If all of your employees can’t trust the advice and recommendations they’re getting, you’re just adding more inequity.

Operating without visibility

You might have heard AI referred to as a “black box,” meaning employers can’t figure out how an AI benefits tool arrives at its recommendations. They’re left to trust a vendor’s algorithm and take their word for it, without any visibility into how it really works.

The opacity of the AI algorithms doesn’t just make it hard to understand why a particular determination was made—it makes decisions hard to challenge. If an employee disputes a denial, “the algorithm decided” isn’t an answer you can give. But, without visibility into the system, it might be the only one you have.

Employees feel this uncertainty, too. When they don’t understand how their data is being collected or used, they’re less likely to trust or engage with AI tools at all. In fact, data privacy is the most commonly cited AI concern among employees, with 38% of employees across job functions saying they worry about it.

Legislation is starting to catch up here. Texas, Arizona, and Maryland have all passed laws that require human oversight before an AI system can issue an adverse coverage determination. Other states will likely follow. But, either way, employers should be asking hard questions of their AI vendors before implementing any tool.

AI in benefits administration: Supporting (not substituting) human judgment

AI is already having a major impact, and there’s no putting the toothpaste back in the tube. It’s here, and it’s here to stay. 89% of HR leaders expect AI to reshape jobs in 2026. That includes the roles of HR professionals, too.

But for all of the buzz about AI replacing humans, that’s likely not how we’ll see it play out in benefits administration. AI will be a valuable tool—not a total replacement for human skills and judgment.

AI platforms can handle the transactional tasks, so the people behind the benefits program can focus their time and energy on the decisions that actually require their expertise. That’s particularly important for complex or high-stakes care decisions like navigating a cancer diagnosis, managing a chronic condition, or figuring out the right surgical option. Those aren’t problems an algorithm should be addressing alone.

Ultimately, the employers who get the most out of AI in benefits are the ones who can identify what the technology is good at while also being clear-eyed about what humans still need to do. AI can speed up the process, but it still takes a real person to know when it’s time to slow down.