The weird and wonderful future of AI chatbots

What is the future of chatbots?

Miscommunication, misunderstanding, downright frustration – at the moment AI chatbots mostly aren’t very good. Yet they’re constantly improving and the potential is huge. However, will the psychological hurdle of talking to a robot continue to prove a barrier?

“A recent mental health project we did using AI powered chat,” Pete Trainor, Founder of human centred design company, Nexus, tells us “had 190 volunteers studied using the bot…”

This is the mostly untalked about side to chatbots. As he explains, half of patients included in the study were told the bot was controlled by a human, like a puppet. The other half were told the bot was fully-automated, with no human on the other end.

“The volunteers who thought they were talking to a computer displayed emotions like sadness more intensely they also said they felt less afraid to disclose personal details about themselves than those who talked with the supposedly human controlled program.”

This is particularly helpful with suicidal men, he says, who struggle to talk openly and share their feelings. “People feel less judged by the computer. Empathetic listening draws people out to make them say more things, plus a feeling of anonymity.”

The standard image of chatbots is in the field of customer service where specific tasks need completing as quickly as possible. And while most of the people I spoke to in the course of researching this article agreed that customer service offers the most obvious immediate benefit for businesses it is still only part of the picture.

Either way though brands must remain cautious. Rob McFarlane, head of labs at Head London suggests: “We could be inundated with chatbots for every brand, action and activity. This could be good or bad as we’ve seen with apps; the bad hopefully get filtered out and the best ones survive.”

However, one thing is certain, the use of chatbots is on the rise. And this is still the start of a very long journey. As Dr Dadong Wan an Accenture Labs Fellow, who is currently leading Accenture’s Tech Lab in Dublin, puts it: “We see huge potential, certainly in terms of the way humans are able to interact and work with machines.”

While Tom Kelshaw, director of creative technology at Maxus describes it as: “The perfect storm of messaging and mobile primacy plus the expansion of harder AI technologies like deep learning and neural networks that allow for natural language.”

To add some context, recent Forester research showed that smartphones users typically spend 80% of their time in just five apps, one of which is often a chat app. This means it is not surprising this is a popular form of brand engagement. A study by Aspect Software also found that nearly 40% of consumers would rather use self-serve apps for customer enquiries than a phone call.

Yet when social media agency MyClever conducted a recent in-depth consumer research study which looked at the entire state of the bot market the story was mixed. “The amount of bad experiences that respondents reported was quite alarming,” Jim Meadows, head of product tells us.

Chatbots may not be up to scratch yet but there have already been some interesting use cases. Tony Tarquini, European Insurance Director at Pegasystems highlights Microsoft’s Chinese chatbot in China called Xiaoice that has 20m people talking to it. While messenger service Kik – which has 275 million users, including 40% of US teens – launched its own bot shop in April, partnering with 16 brands.

Tom Kelshaw, director of creative technology at Maxus points to Sensay, a concierge service startup, which employs a hybrid of bot and human agents to answer customers’ requests. “To improve relatability, they preface any messages from a bot with a blue emoji. It’s an elegant solution for a challenge that will become much larger,” he says.

“Quartz is arguably the most influential digital publication today. The company recently raised the bar in mobile with a bot that embodies the concept of a ‘conversational’ interface. Both the Quartz app and Vice on the Kik platform break news into a set of short snippets of information, and deliver it the same way a friend or co-worker would - chatting to tell you something interesting,” he adds.

At the more ambitious end of the chatbot movement are self-learning programs like Viv, a cross-platform AI assistant made by the creators of Siri, Jo Allison, consumer behavioural analyst at Canvas8 tells us. “Viv aims to serve as an ‘intelligent interface for everything’, with the ultimate goal of replacing search engines as the main way we find information. Yet there remain serious challenges around self-learning AIs, and some doubt how soon chatbots will be able to effectively carry out such demanding tasks.”

And of course there is Slackbot. Part of Slack, this is a sophisticated bot that has all sorts of commands you build into the interface which allows for groups of people to feed into its AI. Mark Armstrong, VP and MD EMEA at Progress explains: “This means you can have a chatbot shared across office teams that would learn who never responds to emails, or the team member who always forgets about meetings. The bot would then learn to send personalised reminders to the guilty party.”

There is also already talk of payment integration in relevant circles. Ralf Ohlhausen, Business Development Director of e-payments company, PPRO says: “Chatbots themselves could be designed to receive payments. To do this, the platform operators would have to build a kind of chatbot app store to enable users to find appropriate chatbots.”

All this has implications for a range of industries. Manish Gajria, Vice President, Product Management, Expedia Affiliate Network (EAN) brand, Ltd believes the travel industry is leading the way in the use of chatbots. While Dan Taylor, head of systems at Fletchers Solicitors highlights their future relevance in the field of law.

“Recognising the benefits chatbots can provide, forwarding thinking law firms are now exploring how such systems can be used in the legal profession, in particular, looking at their role in reducing time consuming data analysis,” he says.

“These AI systems could be extremely valuable in assisting lawyers to review large volumes of current and past case work at a much faster pace. By utilising the large amounts of legal data available in law firms and in external databases, AI could significantly reduce research time and help lawyers to make quicker, more accurate decisions. By removing these time-consuming elements, lawyers could focus their attention on complex tasks like core legal analysis and provide a better quality of service to clients.”

So, where are we heading with all this? Allison at Canvas8 says: “Though they’re still in their embryonic stages many forecasts suggest our new robotic underlings may eventually rise up and depose not only their app predecessors, but also their flawed human predecessors by taking on the grunt work in customer service.”

Armstrong of Progress believes: “Although the initial splash of hype may have dimmed a little, bots are going to continue to grow. Integrated with other apps, bots reduce the need to download a specialised app to perform tasks – a trend that we’re seeing become more and more popular. If the specialised app landscape is shrinking, bot integration will make the appeal of more widespread apps broader.”

Yet Trainor of Nexus issues a warning: “Many businesses are getting it really wrong because they’ve focused on the technology – which has become relatively cheap and easy to implement – and neglected the psychology behind the dialogs.”

“Although companies like Facebook and IBM to a certain extent would have us believe that this is a technology based initiative, it’s really not. It’s a psychology based initiative and so businesses who invest in the tech, but not the psychology will see their project failed,” he concludes.


Also read:

Uncanny Valley: ‘Emotion’ in robots

“Robot Rights”: The future of robots at work

Friendly avatars: How AI is going mainstream

Chetan Dube talks virtual assistants, the Turing Test & mainstream AI

Humanity will invent new jobs long before AI “steals” them all

Office 2021: Why robots won’t end drudgery or steal our jobs