Amazon is testing the first commercial “neural-network based” chatbot able to produce completely original and natural sounding language.
The retailer is developing two separate neural-network chatbots, one consumer-facing chatbot to answer customer queries and another to help its staff respond to requests more easily and focus on those requiring human judgement.
Most chatbots on the market today run by referring to a vast list of automated responses and are able to recognise simple requests and find the most suitable answer.
Amazon’s new generative chatbots will use true artificial intelligence (AI) and state-of-the-art natural language processing technology to produce completely original dialogue.
READ MORE: Chatbots now preferable to humans
It is understood that Amazon will use this technology to aid its staff to begin with, but roll it out to directly deal with consumers after it’s been properly tested.
Currently Amazon is testing these to deal with returns and refund requests as well as order cancellations, training them using 5 million conversation-response pairs gained from 350,000 past interactions.
“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,” Amazon’s machine learning manager Jared Kramer wrote in a blog post.
“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.”