It feels like you can't go a day without hearing about artificial intelligence, right? It's in our phones, it's suggesting movies, and now, it's quietly working its way into the very core of the insurance industry. And let me tell you, it's not just a passing trend.
When you think about insurance, words like "fast" and "efficient" might not be the first to pop into your head. It’s an industry built on careful calculations, tons of paperwork, and deep analysis. But that's exactly why AI is such a big deal here. It’s all about making things smarter and faster.
In fact, a recent Roots survey found that a whopping 82% of insurance executives are calling AI a top priority. And it’s not just talk. More than 90% of insurance companies are already playing with, testing, or actively using AI in some way. The race is on, and it’s moving so fast that regulators are practically sprinting to keep up.
So, What’s Actually Changing?
When I talk to people in the industry, like Kathleen Birrane, the former Maryland insurance commissioner, the changes seem to fall into two big buckets. She really nailed it when we spoke.
First, you have the small, steady improvements. Think of AI as the ultimate assistant, making existing processes more accurate and efficient. It’s about automating the tedious stuff that bogs everything down, from customer service chats to the initial stages of underwriting.
But then there's the second bucket, and this is where it gets really interesting. Birrane calls this the "completely transformative" side of AI. We’re talking about giving companies the power to do things they could never do before. They can think about risk, price policies, and even create entirely new products in ways that were just impossible a few years ago.
It’s a Marathon, Not a Sprint
Now, here’s the reality check. Even with all this excitement, putting AI into practice is tough. While over 90% of insurers are exploring it, that same Roots survey found that only 22% have actually gotten AI solutions up and running in their day-to-day operations.
I had a chat with Mike Upchurch from Domino Data Lab, and he put it perfectly. He said we’re seeing AI move beyond just helping with back-office tasks and into the real heart of insurance, like pricing risk and sorting through claims.
But here’s his crucial point: it’s not just about building a fancy AI model. The real magic is making sure that model stays accurate over time. He warned about something called "model drift," which is a great term. Imagine you have a perfect recipe, but over time, you stop measuring ingredients precisely. The dish slowly gets worse. AI models can do the same thing, starting to misprice risk if they aren't constantly checked against new information from the market.
Let's Look at a Real-World Example: Your Car
If you want to see the good, the bad, and the complicated of AI in insurance, look no further than telematics.
You've probably heard of it. It’s the tech in your car or on your phone that tracks how you drive—your speed, how hard you brake, the time of day you're on the road, and where you go. It’s all about gathering real-time data.
In the commercial trucking world, this is huge. About 82% of fleet insurers are now using telematics, a big jump from 65% just last year. And the results are pretty stunning. Over 70% of fleets that use this tech (along with driver training) have reported fewer crashes and fewer claims.
As Kathleen Birrane pointed out, this data is probably a better predictor of risk than your age or credit score. It can make car insurance pricing way more accurate and, in many ways, a lot fairer.
But There's a Catch...
Here’s where it gets tricky. Do you know exactly what data your car is collecting and who gets to see it? Most people don't. Critics worry that this data could be biased. For example, if you live in a busy city, you’re probably braking and accelerating a lot more than someone on a quiet country road. An AI might see that as "risky driving," even if you're just sitting in traffic. That's a huge problem to solve.
Navigating the Legal Minefield
With all this new tech comes a whole lot of new rules—and a lot of legal questions. Birrane, who is now a lawyer at DLA Piper focusing on this exact issue, told me that insurance companies really need to be focused on three big things.
1. Having a Clear Rulebook
This is all about governance and risk management. Companies need to have solid, written policies for how they use AI. This isn't just about being responsible; it's about protecting themselves from massive lawsuits down the road. And these rules have to keep changing as the technology and regulations evolve.
2. Figuring Out How to Test for Fairness
This one is incredibly complex. How do you test an AI model to make sure it isn't discriminating against certain groups of people? Even states like Colorado, which passed a law about this back in 2021, are still struggling to figure out the best way to do it. Companies are in a tough spot, trying to create tests that are meaningful without accidentally opening themselves up to even more legal risk.
3. Protecting Customer Data
This is the big one. States are getting much stricter about how companies collect, store, and use our personal information. For insurers using AI, this is critical. They're not just using their own data; they're often buying and combining data from all sorts of places. They have to know exactly where that data came from and what their responsibilities are, especially when it comes to getting your consent.
Don't Forget the Human Touch
With all this talk of data and algorithms, it's easy to lose sight of what insurance is all about: people helping people.
I really like what Jeff Wilcoxon from VIU by HUB had to say on this. He stressed that the industry can't just throw technology at everything and forget the human element.
His philosophy is that the best customer experience comes from augmenting human advisors with technology, not replacing them. They’re exploring how AI can make their advisors more efficient and effective, but they’re determined to keep that human connection at the center of it all. And honestly, I think that’s the right way to look at it. AI is an incredibly powerful tool, but at the end of the day, it's just that—a tool to help us do our jobs better.
