Google Duplex: The what, the ethics, the implications for retail CRM and email

Google AI assistant

Announced on Tuesday at Google I/O developer conference, Google Duplex is a new AI technology that allows Google’s AI-powered digital assistant to conduct a natural conversation with a person over the phone, mimicking the chit-chattiness of human speech as it completes simple daily tasks. In this blog, I’ll describe what Duplex is, and will highlight some potential ethical considerations. Finally, discuss the consequences Google Duplex for retail CRM.

What is it?

Duplex is a dazzling bit of voice and AI technology the company says will come to its Google Assistant later this year. In two audio recordings, Duplex phoned local businesses—a restaurant and a hair salon—and completed conversations with a fantastic degree of simulated humanity. If you haven’t seen a demo of it yet, check out this video first.

In the first recording, a woman calls a hairdressing salon: another woman answers the phone and they both chat with each other for about a minute before they agree a time for the hair appointment. In the second demo, also about a minute, a man calls a restaurant to book a reservation; the woman on the other end has a heavy accent and isn’t offering the best information, so the caller pivots to make a new request.

The surprising bit was that neither of the voices who initiated the calls belonged to a human. They were bots and part of a forthcoming release of Google Assistant. However, they sounded human: They said “Ohh, I got ya’” and “Um” and finished query statements with the raised pitch of a question mark. So for the presentation, they completed tasks that generally fall to us mere mortals, whether that meant making a hair appointment or deciding whether it would be better to walk into a restaurant and just take a gamble on a table.

Duplex combines Natural Language Processing (NLP), deep learning, and text-to-speech technology in one platform. The bit that resonates most, though, is the “natural” bit—the Google engineers have trained the Duplex model to match expectations around latency, like pauses after someone says “Hello?”, and change intonation depending how the conversation flows. In other words, to react the way humans do when speaking on the phone.

For Google, Duplex marks a significant step in natural-sounding, autonomous AI-driven conversations. For the rest of us, it straddles a fine line between being enormously convenient and eerily deceptive. Google still hasn’t launched this feature, which will work in Assistant on phones and compatible smart speakers. The company plans to start testing Duplex publicly this summer. In the meantime, there are at least a few features it needs to consider, including how the Assistant will announce itself to unsuspecting humans on the other end.

The ethics of Turing-level AI

When an AI can interact with people in the way that Google Duplex can, we may need to consider the ethics. The best-known group of questions relating to Artificial Intelligence are Isaac Asimov’s three laws of robotics¹, presented in Runaround, his short story published in 1942:

  • A robot may not injure a human being or, through inaction, allow a human being to come to harm.
  • A robot must obey the orders given it by human beings where such orders would conflict with the First Law.
  • A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.

These rules have formed the basis for nearly all the science fiction about robots, almost none of these are directly relevant now, however, since that short story was published, people have been adding new laws. One of the best-known “fourth law of robotics” was added by Lyuben Dilov, in his novel Icarus’s Way. This fourth law is:

  • A robot must establish its identity as a robot in all cases.

Returning to the Google video, the assistant doesn’t say it’s a bot – it actually has mannerisms which make it seem very human. I believe that’s wrong and people are rightly challenging Google on this. Google has already said they will amend this, but it remains to be seen how exactly. Perhaps I’m cynical but Google does not always communicate their intent clearly. In search, Google does disclose which results are advertisements and which aren’t. However, most ‘non-tech’ people don’t easily see which is an advertisement and which is an organic listing.  We’ll have to see, or possibly hear.

How exactly will Duplex be used?

The examples we’ve seen up to now indicate that Duplex may be used in predefined situations – to make reservations and plan meetings. These illustrations fit the virtual assistant (VA) purpose that matches how Google Assistant is advertised. But if an AI is this good at interacting with customers, of course businesses will want to use it to handle inbound calls as well. For example, for helpdesks and customer care functions.

So how exactly does this AI impact retail and CRM?

Google Duplex may not appear to have an immediate impact on retailing but consider this: if your Google Assistant can have conversations such as this with your hairdresser as well as your preferred restaurant, it can be used in retailing to act as your personal shopper. And because it is an AI, it would understand your personal style and tastes and quickly review thousands of products to find the perfect item in the right size. Or even do the other side of the second phone call in the video recording and take the whole booking/sales system and automate it.

The question then becomes: what if your virtual assistant phones the Duplex-driven system on the other side? Will they use human language to complete the job, or perhaps even haggle to get you an extra 10% off?

Mind the (Expectation) Gap

But if the speech is very humanlike, it might lead users to think that all of the other aspects of the technology are very good as well. For AI personalization to be effective in retail merchandising, it needs lots of readily-available content. For retailers, the product catalog is usually the best source of material. The AI system could synch with your product feed, automatically build a taste profile for each consumer, and Wham!, make the idea of a digital personal shopper a reality.

Next steps

It’s hard to predict when Google Duplex will be utilized on a commercial scale. It may not happen for a few years, but it’s going faster than most people outside the tech bubble realize. And as retailers look for ways to connect more personally with customers, we must consider how to integrate AI efficiently into the business where customers will benefit. That way we will not only be able to improve revenues and profitability, but also use AI as a powerful tool to serve customers better.

{Read more|Welcome to the AI revolution 

¹ Wikipedia: Three laws of robotics

Image Source: Eric Risberg/AP/REX/Shutterstock
Comments are closed.