Quarterly Review | Summer 2025

The use of generative AI for hiring is becoming increasingly common. One new way human resources recruits employees is by using AI tools that scan candidate resumes. And new entrants to the insurance industry are using generative AI tools on their end as well.

Shloka Rasal just started her Canadian insurance career. She received her bachelor's degree in law in her home country of India. And worked from there remotely handling both Canadian and U.S.-based insurance claims for global insurance providers.

Starting her Canadian career was like starting fresh. She came to Canada in January 2023 and graduated from Humber Polytechnic’s insurance management program in December 2024. It took her seven months to land her first insurance job as casualty claims adjuster at Intact Insurance.

Her job search could be compared to a cold-calling exercise with AI as a best practices adviser.

“In the beginning of my job search I wanted to target five applications each day,” Rasal said. “Once I got into the job searching rhythm I understood where I needed to focus. When I got the confidence I reduced my application pace to two a day depending on what fit for my target job.”

She used a website called Jobscan that uses AI to analyze resumes, and applied filters on common job search sites to set up targeted and relevant job search notifications. She used ChatGPT and Co-pilot to help refine her resume and cover letter to tailor to each job and follow up and thank you emails, and analyzed numbers on her own performance to iterate her search. She said these tools are great to streamline processes and track performance — but only ever as an aid, never to solely rely on.

Keri Fraser is chief people officer at Westland Insurance Group, a Canada-wide insurance brokerage with around 3,500 staff. Her team of human resource professionals has six people whose job it is to recruit new employees. She said on average each recruiter handles twenty to thirty roles with each job posting amassing hundreds of applicants. It has been a year since the company started using AI tools, including ones that help sift through job applications. The AI is trained on how recruiters do their job and learns how to filter and surface top candidates.

The company’s aim is to reduce administrative tasks to speed up hiring and hire the best candidates.

“We need more time to see the evolution of the impact but one metric where we already see a difference is average time to fill roles, which has reduced by three days out of the 55 days on average it might take to fill a role,” Fraser said.

On the second point, whether they are recruiting the best candidates, a measurable effect is harder to pinpoint. But anecdotally she said their newer hiring practices lead to top performing employees.

“We’ve just helped with recruiters' efficiency, it saves time but we still leave a lot of the recruiting process to human-to-human interactions,” Fraser said.

Andrew Steen, is president of Berkley Canada, a boutique speciality insurance company in Canada. Its U.S.-based parent is Berkley Insurance Company, a Fortune 500 company listed on the New York stock exchange.

Steen is an experienced leader in the insurance industry and spends some of his time helping those newer to the industry and recent grads find their career footing. He said that anecdotally his network says that AI is helping solve problems — and creating them.

“I’m hearing that it has become exceptionally easy to apply for a myriad of jobs,” Steen said. “That can be positive for the candidate experience and it has meant that employers are often faced with being sent hundreds if not thousands of applications for every open role that they have.”

Rasal said that she heard of her peers taking a ‘spray and pray’ approach to job searching: sending applications blindly without looking deeply at a job description or how it matches their skills.

“AI generates what you want it to; first you have to understand what you want out of it,” Rasal said. “The effectiveness of these tools relies on the quality of the initial prompt. You can optimize your search and personalize it and that is what makes a difference.”

Steen echoed Fraser’s point that human resources teams are contending with high levels of job applications currently. He said one of Berkley’s latest job postings drew over 800 applications for one position.

“That’s overwhelming for any organization and particularly overwhelming for small ones,” Steen said. “The most success that we’ve had finding candidates who are really good fits for the role in our company is typically through the network of our people.”

Rasal said meeting people in person, like networking and seeking mentorship, rather than just focusing on online applications, helped her in the job search process. Not just to meet people who are hiring or could be a referral but also as preparation for job interviews.

“Yes, use AI to help with applications but during the interview process it’s not just about the technical skills you have on your resume, employers want to get to know you as a person, and how you handle and respond to certain situations.”

Fraser said given how new Generative AI is there is still a learning curve, including trusting it to do the job well.

“There is still a reliance on humans — the system is learning based on how humans are using it and training it so it can learn properly,” she said. “With AI and other technology the human aspect becomes even more important. Humans understand context and have empathy but there are ways AI can help us be more efficient and effective.”

Fraser added that while there are many companies that use AI in the interviewing process, Westland Insurance has decided to reserve interviewing for humans.

Rasal said that all the companies she interviewed with used video recording for the first round of interviews.

She said generally the process included a company-provided video link that expires within a certain timeframe. Once the candidate is ready to record, a question pops up on the screen and there is a time limit to provide an answer. And it clearly states that the interviewee has to look at the screen and camera.

Steen said that there are reasons to be cautious in using AI during the interview process. Insurance is a relationship-driven business and given the novelty of these tools, AI might not be giving everyone a fair chance.

“We need to be bringing people into our organization that are very good at building relationships for the long term and I don’t see filters helping us with that,” Steen said. “Our business needs thousands of new professionals with those skills, and some of the people that I’m meeting with it’s very obvious that they have those skills but I worry that we are overlooking people based on AI filters.”

He said because of that risk, he directs the newer professionals he meets with to people with whom to have face-to-face conversations.

“I’m convinced that most of them have a place in our business.”

The adoption of Generative AI tools in business is on the rise, and the insurance industry is no exception.

According to a recent Statistics Canada survey, among businesses that reported using AI tools in the last 12 months, 30.6 per cent of finance and insurance businesses were most likely to do so. That sector was also among the highest compared to others in adopting AI tools in the second quarter of 2024. And just over 27 per cent of businesses in finance and insurance planned to adopt AI software in the second quarter of 2025.

Overall the survey suggests that AI adoption is not impacting the number of employees. How that is affecting employee retention, training and ultimately what the return on investment is is another open question.

Colin Young, vice president of building consulting at J.S. Held, a global consultancy firm, said the organization uses Microsoft's AI tools, including Co-pilot. He has used it for editing emails, tracking a busy calendar, transcribing meeting notes to generate action items, and saving time optimizing Excel spreadsheets. He also uses it as a search tool, like Google, for specific location-based information on pricing, for example.

But he emphasizes one big caveat; it’s never used to generate content from scratch.

“These tools help for analytics, but consulting is more than analysis: there is a feeling, an intuition of a problem, that customers rely on that I don’t think AI can comprehend,” Young says. “It’s only as good as the information that it has to work on. Humans can answer questions with more empathy, direction, understanding and information that AI can, at least for now.”

A recent BCG report suggests that with the proper support and training, employees are more likely to use generative AI tools which — if implemented correctly — could improve business performance and make it a worthwhile investment. This requires more than just using AI tools, it is about adopting new processes and workflows. A key challenge is striking the right balance between giving access to these new tools to help employees do their jobs and providing the right training and support.

Nicole Lavigne leads enterprise learning and professional skills development at Definity Insurance Corporation. She oversees onboarding, compliance and mandatory training for over 3,500 employees. Her team, which is under the company’s human resources department, has also created training on how to use AI as a tool on the job.

She said employees are curious about how to best use AI. Definity now has an AI course included in onboarding to cover best practices on how to use it, doing so responsibly, and prompt engineering.

On her own team of seven, they have saved 70 per cent of their time in designing learning materials.

“We can create prompts and iterate and develop courses that have scenarios that are applicable to insurance that brokers, for example, would run into in daily work,” Lavigne said. “It makes what people are learning more relevant and we can develop it much faster.”

Young at J.S. Held said that it’s difficult to calculate investment returns on AI use but, like Lavigne, said it makes a stressful job easier and faster.

“That’s a benefit to our clients because now instead of spending six hours writing a report, it takes two,” he said. “We are busy enough that spending the time saved on doing something else means more gets done. But it’s difficult to measure.”

Beyond how companies can ensure a return on their AI investment, an important question remains: when AI tools are used, how does this affect an employee’s understanding of core technical principles. A June 2025 study from MIT Media Lab suggests that the brain retains information differently when using the AI tool ChatGPT.

The MIT study’s authors balk at the wide-spread commentary that using AI tools is making humans “stupid,” “dumb” or other pejorative terms. Yet the study’s findings — yet to be peer reviewed — does show that brain activity is different when using only the brain versus using AI. So we are saving time on tasks but are we learning the task as effectively as before?

Technical acumen aside, crucial parts of the insurance profession simply cannot be replaced by AI, Young said.

In consultancy he provides services to the insurance industry and gives insights about costs being put forth by restoration contractors and builders to calculate whether the price of work put forth is fair and reasonable.

“Generative AI can’t replace the physical activity of part of what we do. For example, a lot of our consultants physically go on the site to take photos and measurements,” Young said. “There is technology to help with measurements and on site documentation. AI can look at a room and document the contents but how will they see what is in drawers, what’s covered up or if there are serial numbers on contents. I don’t think we are close to AI doing that, at least yet.”

Steen at Berkley Canada said there is fear of the risk associated with using AI tools to complete a process from start to finish. But that is not how the company is — nor intends — to use it.

“Thinking about thousands of different processes at an insurance company, AI is very helpful for accelerating those processes. All with the design to give our customers a better, faster experience,” he said. “It’s just another tool helping humans in how much data they can process so they can make a better risk decision. It’s still the person making that decision. We have not delegated any decision-making to the tool set. At this point we don’t feel that it’s appropriate. It’s an aid, one more tool in our toolbox to help us evaluate risk more thoroughly and faster but it’s still a person with the expertise that is making all the decisions.”

He emphasized that technology tools that enhance productivity are not new. They have been around since he started in insurance in the early 1990s.

“It’s a continuation of those toolsets and we look at it in the same vein: how can this help, and most importantly how can it help deliver services more effectively to customers,” Steen said. “It has a real potential to help improve the day-to-day work experience so we are absolutely looking at that in a very positive way. It’s not about the tools, it’s the people. How do we improve their experience?”

Young said a common way he uses Co-pilot in Excel is to write formulas in order to consolidate data. Before the proliferation of AI, he’d use another recent technological advancement: social media. More specifically, YouTube video tutorials. But compared to Co-pilot, it was time consuming to get the right results.

“Now I get the formulas from Co-pilot and copy and paste into excel that looks at the files and puts everything into one spot,” Young said. “Yes, I have to prompt it a couple of times to get the right formula but it saves hours and hours of work.”

Steen said that at this time in history, people’s values about work and what they want out of life is shifting. Potentially because the Covid-19 pandemic changed how we view work-life balance.

“It’s healthy that people have an opportunity to realign what they want out of life,” he said. “The retention of our people is only as good as we align to their values. And we want to be a place where people do their best, meaningful work.”

Lavigne at Definity calls generative AI a “one start shop and not a one stop shop.” She said the human element of insurance industry work will always be needed, regardless of what aspect of the business people are working in.

“AI just allows us to become a little less tactical by freeing up time and to be more strategic in our work,” Lavigne said. ”We don’t want to lose the human element and for people to blindly trust AI.”

“What fires together, wires together” is a well-known axiom in neuroscience. It suggests that repeated neural activity can strengthen connections and emerge as an established behaviour pattern. Whether this widespread adoption of AI in the workplace is “wiring together” a different kind of employee remains an open question and one that industry is just starting to contend with.