How far can a digital assistant go before people feel uncomfortable? Mark Zuckerberg recently announced plans to create his own AI assistant to manage his personal and professional online interactions. “I’m going to start by exploring what technology is already out there,” he wrote …
Mark Zuckerberg recently announced plans to create his own AI assistant to manage his personal and professional online interactions. “I’m going to start by exploring what technology is already out there,” he wrote.
Zuckerberg does not have to look far for his digital concierge. Website chatbots facilitate customer interaction by assessing what the visitor is looking for and answering questions. Sophisticated examples are already in use in industry.
Royal Dutch Shell has been piloting IPSoft’s cognitive knowledge worker Amelia in its customer service department. Amelia can handle emails and telephone calls. The technology includes artificial intelligence (AI) software that will relate the current interaction to the caller’s previous “conversations”, ie questions, purchases and complaints. By including an emotional intelligence quotient, it knows when to pass a call to the appropriate human colleague.
Mark Brill, mobile strategist and senior lecturer in future media, believes this has great potential for improving customer satisfaction because people find waiting on automated calling systems stressful.
Connecting the dots
No doubt Amelia will have competition as providers have much to gain from the intimacy of this consumer interface – comprehensive user information, connections and preferences. Marcus Mustafa, global head of user experience at DigitasLBi , believes Google is already half-way there.
Digital assistants accessed via personal devices – such as Siri, Cortana and Facebook Messenger’s M – can bring together geolocation and online activities such as booking travel, checking in on social media, finding restaurants or shops. When the user requests a product or service, the DVA will draw up a shortlist based on data analysis. As Brill says: “AI connects the dots.”
“The real value of a virtual assistant on any device is its ability to encompass a wide array of services and apps,” says Martin Hollywood, lead creative technologist at Razorfish London . “Then it listens and from that point it’s mapping data and contextualised machine learning.” As the digital assistant responds to requests, the focus shifts from brand promotion to brands responding to customers’ specific needs and preferences. Hannah Blake, open innovation director for media agency MEC, says: “Brands will have to completely change how they communicate, and move away from mass media messages broadly segmented by gender, age and interests to shorter, more targeted messaging. Natural language processing will be fundamental to digital assistants’ success, and there is a long way to go in this respect.”
One app to rule them all – or a robot middle manager
The digital assistant needs to incorporate user consent, which is critical to any software that collects and uses personal data. Zuckerberg is looking to develop what is effectively a robot middle manager, a single virtual entity that understands and applies his preferences, including consents, approvals and privacy settings, to manage his collection of apps and connected devices.
There are already apps that bring together connected devices. Start-up nCube Home is one example. But unlike apps and connected devices, the digital assistant has a persona. It has a name and gets to know you.
Whereas a corporate chatbot represents a brand, a digital assistant reflects its user’s profile and preferences. “We have not yet determined how brands and their personas will evolve, as digital assistants become a primary means of interacting with technology,” says Adam Falat, user experience practice leader, digital health at Walgreens Boots Alliance. “If you make a request to Siri to refill a prescription, will it facilitate this request end-to-end or act like an operator passing you on to a separate pharmacy digital assistant that will be the face of the brand with its own unique persona?”
The fact that a digital assistant is a potential platform for multiple brands raises opportunities and challenges for advertising and marketing. Hollywood uses the example of a connected fridge that includes the capability to re-order products as they run out. This could create challenges for competitor brands. “The connected fridge could suggest alternatives based on cost and rankings, but to make customers change their habits, brands will need to cut through the noise and remain authentic and relevant.” In the same way as post-purchase marketing messages can annoy customers, brands will need to differentiate between customers’ habits and brand loyalty.
Brands will need to get on digital assistant shortlists, which are basically highly targeted personalised advertising. However, again there are challenges. Blake observes that it will take just one inappropriate suggestion on a shortlist to make a potential customer disengage with a brand. So it’s about making the right suggestion based on user data, analysed with emotionally intelligent AI software . But because of the random elements of human decision-making, the analysis is never complete.
But how far can a digital assistant go before people feel uncomfortable? “At a certain point, humans will no longer be able to tell whether they are interacting with a machine or a real person,” says Falat. “At that point, people will be able to interact with machines just as they do with other people, and may not even know they are doing so.” Blake, however, believes that this breakthrough is a long way away.
Digital assistants that are indistinguishable from humans are at risk of the “uncanny valley” phenomenon – or our natural disengagement with things that appear nearly human. Another consideration is whether emotionally intelligent chatbots and digital assistants require new rules and safeguards.
The main concerns around digital assistants relate to user consent and understanding about the data they share with their own brand (eg Google) and other brands they interact with. Blake flags up associated issues around convergence. For example, if Google powers your digital assistant, will the information it has about you affect your search results? There are times when people don’t want these personalised, for example, when conducting research in a professional capacity. Brands need to recognise that although a degree of personalisation improves service and makes people feel special, the user has to retain control. Otherwise it becomes intrusive and trust is lost.
To get weekly news analysis, job alerts and event notifications direct to your inbox, sign up free for Media & Tech Network membership.
All Guardian Media & Tech Network content is editorially independent except for pieces labelled “Paid for by” – find out more here.
Click here to view full article