‘Human language is the new UI layer, bots are the new apps, and digital assistants are meta apps,’ said Microsoft’s CEO, Satya Nadella, at the firm’s 2016 developer Build conference, using only 18 words to describe a potentially seismic shift in technology’s world order.

It’s the bit in the middle – the bot, an application that performs automated tasks driven by voice or text-based input – which has been generating the most excitement. Facebook made chatbots the star of their developer and entrepreneur F8 conference in April 2016, demonstrating how a user could buy a plane ticket by ‘chatting’ with an airline’s bot inside a Messenger conversation. It then created a marketplace of the 900m people who use Messenger each month by adding a ‘bot store’ to the Facebook platform.

US-IT-INTERNET-FACEBOOK-MESSENGER
Mark Zuckerberg was ambitious and clear: this was the new customer service: ‘You never have to call 1-800-FLOWERS again.’ David Marcus, VP of Messenger, was equally forthright about their impact on consumer purchasing: ‘I guarantee you’re going to spend way more money than you want on this.’ For businesses, it promised a solution with instant scale, algorithmic precision, increased engagement and huge cost savings; for consumers, an always-on, instant access, highly personal form of customer service.

Experience, however, has taught us that every new technology generates as many questions as it does answers. Marcus unwittingly pinpointed the main challenge: ‘We’re conversational creatures … that’s the way our brain functions. That’s the way we’re wired. As a result, it’s probably the most natural interface there is.’

That last sentence should have UX designers fizzing with anticipation: it represents a new field of interaction challenges and the need to develop the skills necessary to capitalise on a new technology with a potential market of more than a billion users.

The first challenge is getting the terminology precise: Nadella mentions bots and digital assistants. The chatbot is, at least for now, relatively limited. It excels in narrow, tightly bounded tasks such as buying a plane ticket or ordering flowers. It identifies key phrases from a user’s input, picks lines from a script, and then performs specific actions from a limited set of possibilities.

A digital assistant is different: think Apple’s Siri, Microsoft’s Cortana or Amazon’s Echo. Deeply integrated at an OS-level, they can leverage data from other apps and the user’s contextual information to fulfil tasks. Compared to bots they can tackle wider, fuzzier tasks such as creating a calendar event, dealing with consequences, adjusting the schedules of those involved, rescheduling other appointments and alerting participants to changes in travel times with traffic updates.

A key difference, therefore, is that brands will have far more control over a chatbot’s behaviour than they will over a digital assistant. They can also add deep specialist skills to the chatbot that a digital assistant won’t have. But there’s a flipside, too: if the chatbox experience is poor, brands risk losing direct engagement with the user, with the OS-level assistant ruling the conversation.

Brands spend millions each year on manifesting their company, values, promises and products; they pay minute attention to the detail of the engagements. A key issue for companies to tackle is how these brand values work with an automated conversational entity.

So how can UX designers help brands get chatbots right? What skills do they have to help their development? What other skills might UX designers need to acquire to earn their place at the table?

The core promise of the bot is, as Marcus points out, ‘a conversation’. These exchanges, particularly between humans and computers, need help. One reviewer described using the current range available in the Facebook store as, ‘like trying to talk politics with a toddler … dull conversationalists … most of my chats sound unnatural, punctuated by moments of frustrating silence.’

Microsoft’s experience with Tay also shows that even when the algorithms behind chatbots ‘work’, the experience for the user can be found wanting. Tay was intended to mimic the language patterns of a 19-year-old American girl; it was launched into the internet with the aim of ‘experiment[ing] with and conduct[ing] research on conversational understanding’, with Tay able to learn from her conversation to get progressively ‘smarter’.

tay-repeat-wall
Tay learnt quickly. It took only 16 hours for Tay to go from sweet innocence to denying the Holocaust and extolling the virtues of genocide. As Zoe Quinn pointed out after being insulted by Tay, ‘it’s 2016. If you’re not asking yourself “how could this be used to hurt someone” in your design / engineering process, you’ve failed’.

That holistic, humanistic perspective is hard-wired into most designers’ thinking. Yet much of the conversational design work is currently being done by developers and data scientists. Beyond bringing that mindset, designers will need to learn how to talk with the teams writing the algorithms that govern exchanges and outcomes. Designers can help create the instinctive empathy for the user that some machine-learning specialists may lack. It’s important that UX teams don’t feel overwhelmed by the ‘science’; the foundation they have gained from using quantitative feedback from app / website analytics should help them with ‘data first’ conversations.

Designers should spend more time on linguistic and language skills and on analysing tone-of-voice. Knowing that your service may be accessed by any one of the 900m people using Facebook should make designers more acutely aware of the norms and expectations in a conversation.

Lastly, designers need to take an elevated, holistic view of customer journeys, end-to-end, and help teams work out where the chatbot is an appropriate tool – if at all. There are strong historical precedences for this: when laptops arrived, they would replace desktops; tablets arrived, they would replace laptops; video calls arrived, they would replace voice calls. None of this happened. Instead, these new technologies were absorbed and found the place where they worked best. Design needs to help marketers and engineers get past the ‘bots are the new apps’ hype and understand when a chatbot works – and when it doesn’t.

What happens if design doesn’t earn its seat at the table?

Peter Sondergaard, Gartner’s SVP and global head of Research, gave an insight into the type of thinking that will emerge in his opening keynote of the Gartner Symposium/IPxpo in 2015. ‘Algorithms,’ he said, ‘are where the real value lies… algorithms define action. People will trust software that thinks and acts for them, therefore, we have to get the algorithms right.’

This is a flawed argument. Algorithms do not define action; that action stems from a conversation which, by definition, is a two-way flow. It’s that interaction that we need to get. So unless design earns its place, Sondergaard points to a future where brands risk ceding control to obscure lines of code, nodding blindly as ‘the computer says yes’.

Nadella predicted conversational language as the new UI layer: humans will need to teach computers human language, conversational understanding, and personal preferences so technology can be more helpful in users’ day-to-day lives. Designers pride themselves on their ability to humanise technology, and now need to apply these in new UX contexts and business partnerships. Here’s what UX teams should continue to draw on, and what skills they need to proactively add: Designers need to take an elevated, holistic view of customer journeys, end-to-end, and help teams work out where the chatbot is an appropriate tool – if at all.

Core skills to extend into chatbot development:

– a holistic, humanistic perspective

– articulating the full end-to-end journey across all touchpoints

– an ability to bridge across silos

– an ability to create compelling interaction visions

– helping getting the small cultural nuances right

New skills to add:

– linguistic/language skills

– an ability to understand the basis of Natural Language Processing

– cross-cultural conversation etiquette

– quantitative data fluency

There was a time when design was brought in at the last hour and told to ‘make it pretty’ or ‘add some wow’. Design has fought its way out of that corner by demonstrating the value it adds across a brand’s business activities. A new interface between humans and technology is emerging, and design needs to show that value again before blind faith in the latest technology hype takes over.