Artificial intelligence has become the benefits industry’s favorite answer to a very old problem: employees do not understand what they are buying. Chatbots now walk workers through open enrollment, recommendation engines match plans to family circumstances, and virtual assistants explain deductibles in plain language. The technology is genuinely impressive, and much of it works.
A more important question sits underneath all the hype. Does the use of AI make benefits genuinely simpler, or are we trying to substitute human interaction with ever-increasing sophisticated tools to help people live with complexity that we should have removed in the first place? After decades in vision benefits, I believe the answer determines whether AI becomes a breakthrough for employees or an expensive way to justify continuing to present complicated benefits with entrenched design failures.
The Confusion AI Is Being Asked to Solve
The scale of benefits confusion is well documented. Research from Voya Financial found that roughly a third of American workers do not fully understand any of the benefits they chose at their last open enrollment, a figure that climbs past half among millennials. More recent data suggests little has improved.
Costly mistakes occur when there is a significant gap between confidence and comprehension. The 2026 Selerix Employee Benefits Survey found that nearly half of employees feel confident managing their benefits on their own, but fewer than a quarter say they genuinely understand them. Thirty-five percent regret the choices they made at their last enrollment. When employees pick the wrong plan, skip preventive care they’ve already paid for, or discover coverage limitations when it’s too late to do anything about it, it not only damages their overall satisfaction with their benefit package but also impairs their confidence in the care they received.
The current trend shows employers are adopting AI faster than they might be prepared to. A 2026 WTW survey found that although only about one in five employers currently use AI within their benefits programs, nearly three-quarters plan to embed it within two years. The shift from experimentation to execution is happening now.
AI Can Improve the Experience
AI deserves credit for what it does well. Feeding plan details into a large language model can help translate giant plan documents into plain English when they may originally have looked like Greek. AI doesn’t sleep or take time off, so it’s available at midnight if that’s when employees want answers to routine questions. It can also compare options across things like premiums, copays, and networks in seconds, not hours. For an employee staring at three or four plan tiers, varying deductibles and copays, and a glossary of unfamiliar terms and acronyms, that is meaningful help.
AI is getting better at personalization, which is the other legitimate breakthrough. The same Selerix research mentioned earlier found that employees who received an enrollment experience customized to their particular situation reported satisfaction rates several times higher than those who received generic communication. AI can create specific guidance for an employee’s family size, health history, and budget in ways that static benefit documents and guides never could.
Enhancing and strengthening the administrative backbone of benefits systems is where AI can really shine. AI can help flag billing errors, easily identify utilization trends, and give employers visibility into engagement data they previously guessed at. These are unglamorous improvements, but they compound over time and free human experts to do higher-value work.
The Trust Gap Nobody Should Ignore
Employers, however, are considerably more excited about AI than the employees it is meant to serve. Prudential’s 2026 Benefits & Beyond study found that 83% of employers want to use AI to help workers understand their benefits, yet only 58% of employees say they would use it for that purpose, and just 24% do so today.
That gap is not just fear of technology. Understanding benefits, making the right decisions when given a choice, and knowing how and where to use them may create a sense of uncertainty. These decisions are consequential and personal, and people want to know that the guidance they receive is accurate, unbiased, and accountable to someone. Removing the human contact from the equation can increase this sense of insecurity if there is any belief that the information provided is not timely and accurate.
A chatbot that confidently misstates a coverage detail does more damage than no chatbot at all, because it erodes the very trust the technology was deployed to build.
Where Human Expertise and Clinical Judgment Still Decide Outcomes
AI is exceptional at collecting data, comparing options, and summarizing information. It falls short because it can’t fully grasp the circumstances of a specific employer: the workforce, the culture, the budget realities, and the long-range goals that shape what a benefit is actually for. Access to information is not the same as understanding it from a human perspective.
AI can make utilization data and analysis more available than ever. Utilization, whether low or high, can have nuance that might not be recognized or explained by AI. Low utilization may signal an efficient plan, or it may signal that employees don’t understand their benefits, don’t understand the value of preventative care or aren’t aware of wellness opportunities, so are quietly forgoing care they need. High utilization may look like runaway costs, or it may reflect strong preventive engagement that actually reduces costs by avoiding expensive problems later. Distinguishing between those readings requires clinical context, experience, and judgment that no dashboard supplies on its own.
The same is true at the individual level. An algorithm can compare two vision plans on premium and network size, but it cannot easily predict whether providers and their staff view a plan favorably, whether patients will be satisfied with the care experience, or whether the benefit delivers value beyond its price tag. In eye care specifically, a routine exam can surface early signs of diabetes, hypertension, and other systemic conditions, and knowing which plan designs actually encourage that kind of care is a clinical question, not a computational one.
Why Trusted Advisors Become More Valuable, Not Less
Every wave of benefits technology has arrived accompanied by predictions that brokers and consultants would become obsolete. Online enrollment, quote engines, and marketplaces all failed to fulfill that prophecy, and AI will too, for a simple reason: as information becomes commoditized, interpretation becomes more valuable. The advisory profession seems to understand this, with recent Council of Insurance Agents & Brokers research showing that 70% of benefits brokerages already have an AI strategy in place.
The broker’s role is shifting from information provider to consultative advisor. AI can generate the comparison; a trusted advisor helps an employer weigh cost against value, evaluate tradeoffs among competing priorities, and choose the option that fits their people rather than the spreadsheet. The lowest-premium plan that produces poor utilization and frustrated employees is not a bargain, and recognizing that requires a human who knows the client. AI also can’t easily recognize intangibles like customer care satisfaction, value adds that some plans and brokers may provide, and the lack of a servant’s heart.
The Question We Should Be Asking: Why Is This So Complicated?
Here is the disturbing truth the industry prefers to avoid. If employees need artificial intelligence to understand their benefits, the benefits remain too complicated. Every dollar spent on navigation technology is, in part, a tax imposed by unnecessary complexity in plan design.
Complexity did not happen by accident. Carriers and intermediaries have created layers of networks, tiers, riders, and carve-outs. These have served as obstacles to care, masquerading as attempts to control healthcare costs, and have accumulated over decades. The reality is that the current system frequently serves the managed care organizations better than the people the plans are supposed to exist to cover. AI that helps employees find their way through this maze is useful when it helps people better access the benefits they have paid for, but it also risks making the maze permanent by making it survivable and might actually allow for more complexity, not less.
The more ambitious goal is to need less navigation, less clarification, less complexity, and more transparency in the design of plans and coverages. Plans built around understandable pricing, straightforward coverage language, and designs that make the right choice the obvious choice should be the rule, not the exception. When a benefit is simple enough that an employee can understand it in one reading, the AI conversation shifts from decoding jargon toward genuinely optimizing care, which is where the technology can shine.
A Framework for Responsible Adoption
For employers and HR leaders considering the use of AI in benefits, three principles are worth carrying into every vendor conversation. First, use AI to eliminate friction, not to excuse it. If a tool exists mainly to explain complexity, attempt to reduce the complexity itself first. Ask carriers for simpler, less complex coverage that encourages preventive services and removes complex hoops employees need to navigate to understand and receive their benefits. Second, keep humans accountable for judgment calls, especially where clinical nuance or employee health is involved. Your employees will be even less happy than they are now if they think an artificial intelligence model is the sole decision maker in the process. Third, measure success by outcomes such as utilization of preventive care and employee satisfaction, not by how many questions a chatbot answered.
There is no question about whether AI will reshape the benefits experience, and much of that change will be for the better. Remembering that technology like AI is a tool, not a strategy, will help ensure the benefits of AI are better realized. The organizations that get this right will pair intelligent tools with sound plan design, human expertise, and a clear understanding of what patients and employees actually need. The goal was never smarter navigation of a broken system. The goal is a system simple enough that people barely need a map.
About the Author
Don Railsback, O.D., is the CEO of Vision Care Direct, a doctor-owned and operated vision benefits organization. He writes and speaks on value-based benefit design, provider-led care models, lowering the costs of healthcare, and the responsible use of technology in healthcare.
