What is AI really and how do I start using it? Mike Fagan, VP Product Management | Jan 25, 2018


“Siri, play Warriors by Imagine Dragons.”  “Alexa, re-order the 35-pound bag of Pro Plan dog food.”

We’ve all become accustomed to these types of interactions. It’s amazing to see just how far technology has come. Interacting with technology is no longer dealing with a screen with a long list of text questions that must be answered in order, and spelled correctly in order to return the desired result.

Artificial Intelligence (AI) is everywhere and it allows people to interact with computers more the way we think rather a pre-programmed way the computer expects. Gartner predicts: “By 2020, customers will manage 85% of their relationship with the enterprise without interacting with a human.” 1

This may lead you to think - Could my customers interact with me that way? What does that look like? Where do I even start? The possibilities make it seem daunting, but these features are becoming more accessible to all of us on a daily basis.

So, what exactly is AI? AI is defined many ways, but in general it is the broad concept that computers can carry out tasks in a manner which is more like humans.  This means they can accept data in a variety of forms and are able to learn from history.

Looking at it in more detail, the first key component of AI is accepting data in a variety of forms. Microsoft calls this Natural Computer Interaction. It’s a collection of capabilities that allow applications to intelligently interpret the world and to naturally engage users. Some example of this are:  

  • Speech Recognition – Recognizing and understanding the human voice. A Stanford News study found that speech recognition is now about three times as fast, on average, as typing on a cell phone.2 Machines can now understand the human voice and in natural sentences. These tools can interpret words, slang, and idioms and understand the intent of the person speaking.
  • Computer Vision – Machines can now understand digital images and video. We’re all familiar with bar codes and QR codes on our phones. These were the pioneers of computer vision. The technology has come a long way from these early roots.  Machines are now able to perform facial recognition, identify topography or landmarks from aerial photographs and can even recognize handwriting on a touch screen or in a picture.  
  • Chatbots – Digital Assistants like Siri, Alexa or Cortana are powered by Chatbot technology. They can accept customer requests, triage severity and make product recommendations all while presenting in a manner consistent with your brand image. You’ve probably already interacted with one of these bots in your daily online interactions.    

The other key aspect of AI is Machine Learning.  Machine Learning is the science of getting computers to act without having to be specifically programmed. Machines learn by mining large volumes of data trying to identify previously unknown patterns leading to development of models that can predict future behavior. Every industry is looking for hidden insights about their customers that can be used to provide a better user experience. These models are at the heart of the recommendation engines that suggest the right product or service based on the customer’s needs and buying situation.

Now you might be thinking – So what should I do next?  How do I get started implementing intelligent applications? Here are two recommendations for you to consider:  

  1. It’s all about the data. Even if you’re not ready to implement any AI tools immediately, start looking at your existing data. These tools rely on the learning that occurs when examining your historical data. What many people find when starting an AI project is that the quality of their current data is poor. Often key pieces of data are missing or incorrect. Your existing data will be the basis of the machine learning models developed. It must to be accurate to develop the proper models and recommendations.

    Be prepared to undertake a Data Cleansing exercise to identify missing or incorrect data so you can begin the process of cleaning up the data. Then implement strong Data Governance procedures to ensure complete, high quality data is maintained going forward.

  2. When you are ready to implement AI, listen to your customers. What are they asking for? What need is currently unmet? Don’t just implement some new technology because it’s the new shiny thing. This shouldn’t be a solution looking for a problem. For instance, if your clients need to be able to speak to you, consider speech recognition options. If pictures or videos contain key information, you might consider computer vision to scan photos your customers submit. Take a long look at your current customer interactions and think about how you want to engage them in the future and carefully determine what new features you want to make available to them. 

Insurers are already using these tools in interesting ways. Some are using chatbots to accept application information and provide the quote options to insureds online. Others are using computer vision and machine learning to handle automobile physical damage claims. The insured submits a photo of the vehicle damage and the system is able to prepare a damage estimate.

We’ve only scratched the surface of the capabilities these tools can provide. In this rapidly changing marketplace, the companies that embrace AI and strategically leverage its benefits will be the ones best positioned to succeed.

Mike Fagan has worked in Property and Casualty insurance and automation since 1988, including ten years with Cigna's Special Risk Facilities. Mike joined Instec in 1996 and works closely with Instec’s clients, and has delivered many of Instec’s new software releases, ensuring a smooth transition for clients having the prior release of the software. He received an MBA from the University of Chicago in 1995, specializing in Finance, and received a BS in Information and Decision Sciences from the University of Illinois.