AI promises to transform benefits admin but adoption tells a different story as ‘HR is trailing behind,’ says Normandin Beaudry
Each month at BPM, we offer a slate of articles and content pieces that go deep on a particular topic. This month, we're focusing on consultants and the role they play in shaping benefits, pensions and investment outcomes.
While artificial intelligence is expected to change how Canadian organizations administer employee benefits, recent findings have found adoption has yet to catch up with the hype.
According to data from the MBWL International global benefits survey, which captured responses from roughly 200 Canadian employers, the use of AI in benefits remains thin across most functions.
Sophie Limoges, principal of group benefits at Normandin Beaudry, outlined some of the key findings which will also be presented at an upcoming webinar hosted by the consulting firm next week.
“The pace is accelerating,” she said. "HR will have to pick up the pace because AI investment and AI development in organizations are mostly focused on operations. Executives the C-Suites are now expecting very fast turnaround times with responses from their team and they can respond by the end of the day because AI allows them to track data requests. HR is trailing behind. Right now, HR is not able to have that turnaround so at some point if it's not already an issue, it's going to be an issue.”
According to Limoges, the survey asked employers about seven AI-related benefits tasks; on five of them, between 81 and 91 per cent of respondents reported no usage at all. Personalizing employee communications was the most active category, with 36 per cent of organizations starting to use AI and seven per cent qualifying as heavy users. Enhancing employee engagement and satisfaction came in second. While year-over-year comparisons show the numbers climbing, Limoges couldn’t put a timeline on broader adoption.
Still, Limoges underscored personalization in benefits communication comes down to timing, particularly as employees are already overwhelmed with messages, and generic blasts miss the mark. She noted some organizations are starting to use demographic and age data to target messages to the right populations at the right moment, and insurers are doing the same.
"AI won’t solve everything. If you want it to be efficient, it has to be at the right moment," she said.
Limoges anticipates the pace will pick up as organizations feel mounting pressure from advancements elsewhere in the business. When the survey asked employers what they most want AI to accomplish in benefits management, cutting admin burden for HR came in first.
Communication followed close behind and also kept resurfacing across nearly every topic in the findings. Employers not only see it as a tool for speed and reach, but also to manage costs by helping plan members make smarter decisions, like choosing the right approach to claiming prescription drugs.
Meanwhile, predictive analytics is the least adopted AI function in the survey. Limoges said forecasting future costs ranked as the most unused task they measured, with 91 per cent of organizations reporting no AI use at all and just two per cent qualifying as heavy users.
"They really seem to be focusing more on the employee experience than doing those analytics for AI," she said.
As for the challenges in broader AI adoption, Limoges acknowledged 70 per cent of organizations believe data privacy and security to be the dominant concern. Employers are asking basic questions about where their information lives and who can access it, particularly given Canada's regulatory landscape.
After privacy, the next tier of challenges appears relatively close together. Integration with existing systems ranked high, as most organizations already run established HRIS or payroll platforms and struggle to see how AI fits alongside them. Cost followed as the third-largest barrier. Notably, lack of senior management buy-in landed near the bottom, which Limoges suggests the resistance isn't coming from the top but from the practical realities of budgets, infrastructure, and competing demands on the ground.
“I don't think people feel that it's really a top-down issue. It’s more about how you work with your current environment and how you actually invest in those new solutions,” she noted. "There are always opportunities for optimization and the objective is always to be able to offer more at the end to the employees at a better cost.”
Among Normandin Beaudry’s clientele, Limoges acknowledged the most common move right now is building internal chatbots - often using tools like Microsoft’s Copilot - to field routine questions about benefits policies and reduce the volume of inquiries hitting HR teams.
Limoges described it as one of the more accessible entry points for AI, one that doesn't require a full system overhaul. But accessible doesn't mean risk-free. She said chatbots need to be stress-tested thoroughly before going live, noting that her team has sat down with clients and deliberately fed their bots absurd questions and notably received genuinely problematic answers.
While she emphasized she’s not an AI expert by any means, she noted the technology can't be built in an afternoon and left unchecked. It requires guardrails, limits on what it can recommend, and repeated testing to ensure it doesn't steer employees toward inappropriate advice, particularly on sensitive topics like retirement savings.
“You don't have to invest in a full system because it's fairly accessible,” Limoges noted.
Technology is valuable for surfacing data. Analyzing claims, surveying employees, identifying where plans can be improved, but the decisions that follow still require people, Limoges said. She said plan design is built on “guiding principles”, employer values, and an organization's employee value proposition, none of which should “be dictated by a machine,” she said.
While Limoges acknowledged AI can help map the strategy and identify which messages can reach which groups, she believes the last mile ultimately still depends on managers, HR road shows, and benefits town halls where employees show up and ask questions in person.
Even at organizations that have invested heavily in technology platforms, employee feedback has been clear: people still want the option to talk to someone. Younger workers may be more comfortable navigating systems on their own, but the demand for a human touchpoint persists across demographics.
"I feel that the human side will always be important in some way," she said.
