When Google introduced its Duplex AI service at the company’s recent conference it took a lot of people by surprise as it seemed like such a major leap forward.
Google Duplex’s intelligent blending of deep learning, natural language understanding, speech recognition and text-to-speech (TTS) technologies creates a virtual assistant that is so realistic that it is really hard to distinguish whether its speech output is a human or not.
Specifically, Duplex draws on Google’s latest TTS developments to add speech disfluencies – all the many ums and ahs and artificial pauses that combine to make human speech so distinctive – to its dialogue. However, Google also challenged the status quo by demonstrating how its new natural-sounding assistant could also be used to initiate bot calls that could be executed using Google Assistant and draw on data from back-end systems.
It was this reversal of the traditional bot model – where a human contacts an organisation and has to engage with a computer via a pre-determined interaction structure – that surprised so many observers. By introducing a virtual assistant that can itself reach out to humans, and that uses speech that’s almost indistinguishable from normal human dialogue, Google was accused of enabling organisations to deceive their customers by making its virtual assistants too human-like.
In fairness, Google has actually addressed this issue head on and says that it’s working hard to get ‘the expectation right’. It also, quite correctly, has highlighted the fact that this new level of heightened virtual assistant capability has the potential to unlock significant value for organisations and their customers alike.
The initial controversy around Google Duplex only serves to confirm that all intelligent assistants, regardless of how intelligent they are, still require deep UX and customer journey expertise if they’re to deliver a consistently excellent experience for customers. However, Google’s disruptive drive is clearly stretching the boundaries of how assistants can reshape customer engagement.
It’s a challenge that Sabio is addressing directly with our Digital Front Door platform which is designed to provide a configurable natural language capability that can operate as the first point of engagement for our customers’ contact centres. We’re currently working on a number of proofs of concept, focusing particularly on how all customer interactions – whether voice or text – can be processed via an intelligent natural language capability before either self-service or human assisted resolution.
Google Duplex has shown just how quickly intelligent assistant technology is evolving. We’re looking forward to introducing this kind of capability into live customer journeys very soon and look forward to helping our customers provide a consistent Digital Front Door experience across all their key channels.