Amazon is testing customer service chatbot technology that could produce the first ‘generative chatbot’ capable of original dialogue in real time.

A blog posted by Jared Kramer, an applied-science manager on Amazon’s customer service tech team, explained that the e-commerce giant is trialling end-to-end neural-network based chatbots.

Based on new natural language processing technology, the project marks a definitive shift away from traditional customer service chatbots, that typically have a wide range of pre-scripted response to prompt words and key terms within a customer’s enquiry which are fed into ‘flow charts’ by automated agents.

Amazon’s chatbot instead uses huge training data sets and predictive text to come up with realistic-sounding copy or dialogue.

Kramer said that the two neural-network based systems would help customer service agents in new regions to handle common customer service requests automatically while the other helps them respond to customers more easily.

“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,” he said.

Randomised trials comparing results of the new neural network chatbot systems to the rules-based models have shown that the new methods “significantly outperform” the other one, Kramer stated.

“It is difficult to determine what types of conversational models other customer service systems are running, but we are unaware of any announced deployments of end-to-end, neural-network-based dialogue models like ours,” he commented. “And we are working continually to expand the breadth and complexity of the conversations our models can engage in, to make customer service queries as efficient as possible for our customers.”

Share Story: