The chatbot revolution has begun – but there’s no substitute for the human touch

Having a legion of robots around to do our bidding and make our lives easier has been a sci-fi vision for decades, but in recent years it’s finally become a reality. Not only do we have physical automatons performing tasks, but in recent months the concept of artificial intelligence has finally moved beyond being a pipedream with hundreds of AI-powered chatbots debuting this year.

The rapid pace of development means that service desks should begin seriously considering how they can integrate chatbots into their strategies.

The first wave of chatbots

Facebook perhaps deserves much of the credit for the AI helper becoming a reality, with Facebook AI Research (FAIR) helping to bring the concept of the chatbot to the mainstream. Facebook’s Messenger platform, which launched in April, introduced a wide variety of chatbots able to respond to users and learn their needs.

The first wave of chatbots cover a wide range of services, particularly when it comes to their obvious application in answering questions and providing information. One example is Sure, a chatbot that offers guides to restaurants and cafes, answering questions and even recommending its favourite dishes.

Many of these are also experimenting with injecting personality to make them more engaging. Hi Poncho, for example, takes on the guise of a cute cartoon cat that answers users’ questions about the weather and can provide customised daily forecasts. Meanwhile, the Operator chatbot is an example is a shopping assistant specialising in recommending products and offers. Retail is widely expected to be one of the most popular applications for chatbots, and one of the sectors where it will have the biggest impact on customer service.

Alongside tech giants like Facebook and Google, AI has become the new goal of the latest crop of students and start-ups. Other recent examples include MyKai, a banking application which tells you how much you have spent on different things and is able to learn your habits. Meanwhile, a 19-year-old student recently created a free service called DoNotPay, a 'chatbot lawyer' that can tell you if you are actually liable for parking fines you have received.

Glitches in the system

With so many applications already being explored at every level of the tech industry, it’s clear that AI is well on its way to becoming mainstream, especially when it comes to customer-facing chatbots. However, as with all new technology chatbots are far from perfect.

A good example if Microsoft’s Twitter chatbot ‘Tay’, which was designed to tweet based on information it absorbed from other users… but was shut down after it started posting racist comments. Clearly there are a lot of potential bugs to work out, especially when there’s a broad remit for the AI to interact and learn within.

Indeed, scope is one of the major limitations for the current generation of AI. Poncho and other chatbots have faced mixed reviews due to the narrow focus of the language they are able to recognise and respond to. This means that, for now at least, there is a high likelihood of them being unable to understand or interpret requests, which will quickly become frustrating for the user.

Introducing AI to the service desk

Much like physical robots at work in factories, AI are much better at simple, repetitive tasks. When they do perform correctly, they can do it far more quickly than a human and will never get bored of mundane tasks. Service desks should look at deploying chatbots as a first line of contact to handle simple requests via self-service facilities. This will free up staff to handle more complicated and high-value demands, providing an opportunity to move the human workforce to more strategic roles elsewhere in the supply chain.

Service desks should also definitely have human staff ready to take over if the bot has any issues, or the request is too complicated for it to fulfil. Chatbots which are handling requests that would normally be dealt with by a human also definitely need to be fully integrated into the rest of the service desk system. Requests must be accurately logged and prioritised in the same way as other contact points so that complex issues and requests can be progressed and service level agreements fulfilled.

For now there is no substitute for the human touch and it will be a few more years for the technology to be deployed and gremlins to be worked out.

However, both consumer and business-facing service teams should be prepared to embrace the idea of AI representatives. All industries should begin looking to get ahead of the curve by investigating ways they can integrate the technology into their customer service alongside their human workforce.

Neil Penny, Product Director at Sunrise Software