Capital One Managing Product VP on Building Smarter AI Products (PART 2)

When does it make sense to apply AI/machine learning (ML) into a new product? For any problem you’re trying to solve, first identify if any ML algorithms already exist in similar scenarios. Looking at chatbots, you have algorithms such as NLP, NLU and NLG are already out there. With many AI products, you already have a precedent to work from.

Now, let’s say you wanted to build an intelligent assistant. The first step is to define what kind of questions you want to answer. Will they be specific or general? Next, determine if you have enough data to train your model. If you don’t have enough examples, you’ll have to build it. This can lead to limitations, however, based on sample set richness and diversity.

Staffing Is Critical

If you are truly building something from scratch, your company should have the staff who can assess the applicability of machine learning to your business question. If you don’t have experienced data scientists or ML engineers, consult with one to assess your project before building. Otherwise, it might just become one never-ending research project.

Existing AI/ML Solutions

If you can clearly articulate a small group of rules to make a decision, then ML is overkill. The fields where you can find already established AI/ML tools include image recognition, NLP and anomaly detection.

Anomaly detection works when you have a lot of patterns and data and a constant data stream, like in a machine or system with sensors. Here, machine learning can detect when things go wrong and quickly trace back to the culprit event. In this case, you end up with a predictive model that can prevent machine or system-based breakdowns.

Product Teams Count Early

For AI feasibility testing, measure the amount of data and asses any already available algorithms first. Then begin with data and feature engineering to start building the model. It’s important to get product people involved early to make sure you’re solving the right problem with the right design. Product teams help make sure the end user needs are truly addressed, since machine learning algorithms live hidden behind the interface.

 

Click here for Part 1

Click here for Part 3

About the Speaker
Margaret Mayer
Capital One Managing VP
Margaret Mayer is the Managing VP of Messaging and Conversational AI at Capital One - setting the technology strategy for the company's banking chatbot (Eno) and enterprise-level messaging platforms. Over the past two decades, Margaret has held several leadership positions at Capital One - including software engineering, consumer identity and partnership development. Margaret holds a PhD in industrial engineering from Lehigh University and a bachelors degree from Cornell University. She currently lives in Richmond, VA.

Recent Posts