Amazon has revealed that it is currently trialing two AI-based systems to handle shopper inquiries in an effort to improve the customer service of its ecommerce platform.
While one of the systems is able to field requests from customers automatically without human intervention, the other is designed to help human service agents respond to inquiries faster and more easily.
Applied science manager at Amazon’s customer service technical management organization, Jared Kramer explained in a blog post how the ecommerce giant’s automated agents use neural networks to make decisions as opposed to just rules, saying:
“Most text-based online customer service systems feature automated agents that can handle simple requests. Typically, these agents are governed by rules, rather like flow charts that specify responses to particular customer inputs. If the automated agent can’t handle a request, it refers the request to a human customer service representative. On amazon.com, we’ve started phasing in automated agents that use neural networks rather than rules. These agents can handle a broader range of interactions with better results, allowing our customer service representatives to focus on tasks that depend more on human judgment.”
According to Amazon, it’s customer-facing AI-based system is using a template ranker to control the automated agent’s vocabulary. This involves the AI model choosing from hand-authored response templates but the company plans to begin testing out a generative model where responses are crafted on the fly.
The templates are made up of general forms of sentences with variables for things such as product names, dates, delivery times and prices. However, the model is also able to incorporate new templates with out too much additional work because it has been pretrained on a data set of interactions between customers and representatives.
Amazon’s researchers trained separate versions of both AI-based systems for return refund status requests and order cancellations. The order cancellation model receives some information about a customer’s account profile as well as dialogue context. The response ranker also uses an attention mechanism to determine which words were particularly useful for ranking a response.
Amazon isn’t the only company using AI to better deal with customers and the customer experience management firm Servion has predicted that AI will be used to power 95 percent of all customer interactions by 2025.