Interview: Gia Sawko, Director of Claims, on How AI Helps Make Claims More Efficient and Effective

Gradient AI’s Marketing team recently sat down with Gia Sawko, Director of Claims, to discuss how AI is being used by insurers in their P&C claims operations, and the impact it is having. In just a short conversation, Gia shed a lot of light on artificial intelligence and its expanding role in P&C claims management.

QUESTION:

At a high level, can you describe how insurance claims professionals benefit from the use of AI?

In the most recent conversations, when I've been asked the question of the value to the insurer in how claims and AI work together, or where it's going to go, I think about a lot of the opportunities to change the workflow.

Had I been an adjuster 30 years ago, looking at what people are doing today in claims, it doesn't seem all that different a process. Insurance claims themselves have become more sophisticated. And so, for carriers, the job needs to meet the complexity of the space where we're in today. That's really what AI and machine learning can do.

My favorite part about it is that now there's a way to use data for adjusters and supervisors that they didn't have before. If they were taught to do everything by instinct, they learned by repetition. Times have changed, and now, when they interact with the model and the data, especially Gradient AI’s data -- because it's such a deep data lake – the adjuster for the first time has this visual, this tool that they can interact with, to either validate their reserves and their position, or to reconsider not only what a claim is going to cost, but what actions they need to take. It also increases the accuracy of what they're predicting.

AI is using the same best practices as seasoned adjusters. It's built on the same Law of Large Numbers as when an adjuster was trained 30 years ago. We talk a lot about this in claims: ‘if it walks like a duck, swims like a duck, it's probably a duck.’ And that's where that instinct had come in, because you could trust by just the sheer number or volume of claims that you knew how it would end or its outcome. For example, this claim that was unwitnessed on a different location after the person was written up, and they've been at work for 6 weeks. Those are all red flags. That claim was going to cost a lot of money. It had a high probability of litigation, and perhaps a lot of intent to sue or to retain counsel.

These same patterns exist in claims, and the AI model can learn that. It really is like a parallel universe between the traditional approach and AI, but because AI is faster, it just gives the adjuster or their supervisor information more quickly. I don't think it's better, I think it's faster. A very good adjuster is a very good adjuster, because they've learned how to interact with a claimant based on what they see. Even a good adjuster can benefit from using the model, and a new adjuster can definitely benefit from using the model. They just don’t have that kind of depth of experience, and they're not given the same length of time to be exposed to so many claims. So, it's the perfect time for the industry to embrace machine learning and AI, as the next generation of adjusters comes in.

QUESTION:

In terms of the talent gap in the insurance industry, the labor shortage, and all of those challenges, can you just talk a little bit more about that and the potential benefits for new adjusters coming in?

You know, this talent gap has come up in conferences throughout this past year. Everyone is wondering, “What will we do with the insurance industry given the labor challenges?” In the last 6 weeks, I’ve evolved my vision on it.

Adjusters that have been doing claims for a really long time probably have the slowest adoption of using the models just because it's such an innate nature for them. Back to that instinct, they're really good at what they do. Because of that, they feel like AI doesn’t necessarily help so much. It's kind of like telling them something that I already know.

But as these adjusters are starting to retire, then I think about that next generation. I think about attorneys that are in their mid to late twenties or people coming out of school. They're used to being with data. They're used to interacting with that kind of technology. And I feel like the old way of doing it is not an attractive job anymore, because it is so archaic. Now, as the industry evolves, it becomes “Hey! Come, join us! We're the fastest growing tech industry to be a part of!” I think now, we're going to actually flood the market with people who want to come here, use AI, and do all the noble things that adjusters and supervisors have been known for.

QUESTION:

And what about like the benefits of the guardrails or guide posts that AI provides?

Yes, that that's even more of a reason to use AI, because before it was the supervisor that would read your diary and tap you on the shoulder and say, “Hey, when you see this pattern, you already know that something has a high probability of happening.” Now with AI, the neural networks do that for you, they recognize the pattern. AI gives this single thread of a best practice.

And the other thing I always find interesting is that a lot of people could be good at claims, but not everyone can train someone else how to handle claims. They know how they do it, but either they're not articulate, or they just don't have the experience teaching someone else. So, these guardrails that AI provides kind of become that gentle teacher, because it's stopping you and saying, “Hey, consider this file for an IME, or consider this file as costly, so you need to be more in touch with that injured worker.” In this way, it's like really a tool to help guide newer adjusters.

QUESTION:

Do you think we're moving to a time when claims adjusters won't have to touch every claims file every day?

I hope so. That is my dream, you know. I tell this story a lot that as an adjuster, if I had 14 claims on diary, I took them as 1, 2, 3, 4, 5, 6, 7, all the way to 14, and if that fourteenth claim was the worst claim, hopefully, I got to it because chances are I was interrupted by the phone ringing, or some other distraction came up. Now with AI, you can scramble that deck, look at those priority claims and take action. It doesn't mean you'd ignore the claims that are on track, but maybe you wouldn't need to look at them at the same frequency. I sure hope we’re moving towards that.

QUESTION:

When it comes to AI, what’s the value of the data, and what does it mean to have access to a vast data lake? How does that impact insurers?

I'll use Whole Foods as my example. When I was a Managing Director there, I knew my claims and my transactions. How many reserve changes, how many new claims, how many claims that we closed. I knew everything there was to know about Whole Foods’ claims, and I knew everything there was to know about best practices, because I've been doing it for a long time.

But what I didn't know was how well Whole Foods did against other companies. And my claims were very specific to retail and food and hospitality. Where would I go to find any other claims to benchmark? I could call my friends, you know, at my competitors, but that would be weird. I could call other retailers, but they weren't the same business. So I really could never leverage anyone's data but my own. And so now, having a deep data lake, and what I love about the data lake here is that it's anonymized, it's de-identified, the data is completely private and secure when it goes into the data lake, it's stripped of all of identifiable information. We have no idea what claim it is from. Any single claim in the data lake is just one of our claims. So now, you have access to this deep data lake, and you have the opportunity for those comparisons to learn more about how your claims compare.

Also, you can learn about the behavior of claims in other states or regions, or across the country, and in terms of moving into a new market or expanding geographically one of the great things about having a contributory set is if you’re a company with a national footprint, or you’re a carrier that’s writing some policies in New Mexico, and your adjusters are handling multiple claims but they don’t have that many in one particular geography to compare a given claim to, or they want to expand into a new industry – to serve retail clients or restaurants, or construction, or office workers. For me as an adjuster, those meant very different injuries. And I think if a carrier was going to enter into a new market, the ability to see that data or learn about that market would be really helpful.

Explore the benefits of Gradient AI's P&C or Workers' Comp Solutions. 

 
Schedule a Demo →