Feb 54 min read

InsurTech Ohio Spotlight with Chris Luiz

Chris Luiz is the Cleveland Founder at InsurTech Ohio and the Director of Solutions Architecture and Customer Success at Monitaur, an AI governance software platform helping companies build, manage and automate responsible and ethical governance across consequential modeling systems. Chris was interviewed by Michael Fiedel, Co-Founder at InsurTech Ohio and Co-Founder at PolicyFly, Inc.

Chris, define ChatGPT and Generative AI.

“Generative AI models use neural networks to create new content similar to the content of the training data. Language in, language out, images in, images out, etc. ChatGPT is one such example that has taken the world by storm.

I think ChatGPT's impact on the insurance industry cannot be understated. It absolutely dominated the technology conversation, and now every carrier is trying to solve how they use Generative AI. This is a more important opportunity than simply focusing on the hype around ChatGPT alone.”

Technology Impact Overview

The impact overview is a series of scores from 1 (not at all significant) to 10 (game-changing) of Chris' perspective regarding the potential value and ongoing adoption of Generative AI technologies.

Realized Value - Score: 2 / 10 (Barely Significant)

How impactful has Generative AI been in the last year?

“I’d say nine out of ten from a hype perspective but only two out of ten from a delivered-value perspective. There’s a lot of positive conversation but not a lot of value realization.”

Potential Value - Score: 9 / 10 (Incredibly Significant)

How would you rate the potential future impact of Generative AI?

“If we were only rating ChatGPT, I’d say it’s pretty low. But, the potential of Generative AI would be a solid nine. Our industry will derive immense value from these technologies over time.”

Current Adoption - Score: 2 / 10 (Barely Significant)

How broadly was Generative AI adopted by the industry last year?

“At most, I’d give it a two. Last year, we saw corporate flailing, FOMO-driven purchases and some early development of models over existing corpora. The big carriers likely have a few models in or are moving quickly to production, but they are addressing a tiny fraction of the problem space.”

Velocity of Adoption - Score: 6 / 10 (Very Significant)

How fast do you think we will see Generative AI as a commonplace technology within the industry?

“I would say six. I expect nearly every carrier will be exploring it for something shortly, but full penetration is going to take years.”

Chris, share your overview of how you see Generative AI playing out from here.

“The single, most impactful technical advancement of 2023 was the release of Open AI’s ChatGPT and the mainstreaming of the Generative AI (“Gen AI”) discussion it provoked. This applies across industries. These types of models are going to change the world, and our industry, in many ways.

Language transformer models have been around for some time, doing similar though less broadly-scoped tasks, but the progress and impact of these models were largely invisible to the general public. The release of Open AI’s GPT 3.5 model behind a simple web interface marked a transformative moment in artificial intelligence and reshaped the world's perception of what is achievable with AI. It also created enterprise-level FOMO like never before.

ChatGPT demonstrated unprecedented language generation capabilities, enabling mostly natural and mostly-coherent interactions with users. It immediately changed how people engage with AI and also sparked discussions on ethical considerations, potential applications and the future of human-AI collaboration. But, focusing on the ‘chat’ component of large language models with regard to our industry is burying the lead. The true value of these models will be realized when the synthesis of vast amounts of unstructured data leads to high-dollar decisions.

In the insurance industry, we produce and consume voluminous amounts of unstructured data in the form of claims notes, customer call transcripts and property and risk reports, amongst other sources. We use them to make decisions that impact people, drive our underwriting profits and instigate fraud investigations.

These models can provide detailed explanations and summarizations and offer personalized assistance across all of these areas. The insurance industry will eventually use language models in nearly every place where the summarization of language is valuable, within carriers and across the value chain. These modes and their adaptability will lead to more seamless and efficient administrative employee and customer experiences.

As conversations around the power and uses of language models unfolded, concerns about responsible AI development, bias mitigation and the ethical implications of AI systems became board-level issues. The release prompted a global dialogue on the responsible use of AI and the need for comprehensive guidelines to ensure their ethical deployment.

ChatGPT not only expanded the possibilities of AI but also catalyzed a broader societal reflection on the role of AI in shaping the future. That conversation for AI regulation is still nascent, and strong governance practices are not yet widespread. We’re still at the ‘thinking about it and talking about it’ stage versus operationally mature. We have a long way to go.

Personally, I am very interested to see how the question of copyrighted training data is addressed and to understand what it means to deploy a secure, customer-facing language model. Especially in regulated industries like our own, where the question of data and model ownership and privacy have not been adequately addressed.

Overall, I am hopefully optimistic that the world, and our industry, will learn to use language models in a productive way that drives real ROI and customer and employee satisfaction.”