In our latest installment of “Questions From Our Last Meeting,” we cover a common question that we receive from financial institutions who are already familiar with some of the building blocks of artificial intelligence (AI) – in this case, Machine Learning.
Setting the Scene
Date: January 2022
Financial Institution (FI) Type: Credit Union
Size: $5-10 billion in assets; > 250,000 Members
Question: How does your Machine Learning model leverage missed utterances to improve accuracy?
- This is a largely debated topic within the Natural Language Processing/AI community. Ultimately, we are delivering an experience for your members, and what works for one member may not work as well for another. As part of the managed Virtual Financial Assistant (VFA) solutions that we offer, when utterances are missed (defined by something the AI was not trained to manage/identify) they are flagged in our system for both programmatic review and human intervention. The updates happen automatically without the credit union needing to do anything (assuming we do not need you to opine on a change).
- Abe.ai has built algorithms that group missed utterances to review with clients on a regular basis. This exercise is used to determine the roadmap for the VFA and is also where we are able to proactively alert you of issues and opportunities. If, for example, a high frequency of utterances are complaining about failed check image recognition, we will proactively reach out and let you know there is likely something going on with the Remote Deposit Capture feature.
- Our AI is not 100% artificial – there is a deep, experienced team of product specialists, engineers, machine learning experts, data scientists and more that are involved in training, enhancing and expanding our products to provide truly intuitive solutions.
If you’re interested in learning more, reach out to us here for more information or to request a no-obligation demo of our multi-tier Virtual Financial Assistant.