AUTUMN//WINTER 2018 41 40 AUTUMN//WINTER 2018 PEOPLE tend to regard artificial intelligence (AI) either as friendly androids, like the chirpy robot double act C-3PO and R2-D2 in Star Wars, or monsters like the psycho killer HAL in 2001: a space odyssey. Sadly, AI in reality is much less glamorous. Simply put, it is a collection of software tools designed to work invisibly in corporate data systems to bring human-like skills to areas where today’s computers are lacking. Technically, AI is all about pattern recognition, the skill that we humans find instinctive but until recently has been a major challenge for computers. Now, however, machines can scan digital images and audio and even video for patterns that indicate such things as faces, objects, handwriting and spoken words. They can identify them with huge accuracy, as well as understand what they mean. The same process has been used to understand language and accurately generate spoken or written words. Amazon’s Alexa and Google Assistant are the most prominent examples of AI systems in action, and anyone who has spoken with them will testify to their slightly spooky ability to understand what you mean and respond appropriately (though both are still capable of being frustratingly dim or hilariously wrong.) The success of these automata shows that people are increasingly willing to chat with robots when shopping online, according to Anne de Kerckhove, CEO of Freespee, a Swedish company specialising in online customer contact software. “According to PwC’s Global Insights Survey 2018, which surveyed more than 22,000 consumers worldwide, customers are happy to talk to robots,” says de Kerckhove. “Brands must convince customers that they are authentic and caring in order to thrive – and though it may seem counterintuitive, AI could lie at the heart of this. AI can increase the human element to customer service – allowing businesses to offer the very best of human one-to-one service in an online space. In fact, 60 per cent of respondents in the survey agreed that AI can reduce the time it takes to get answers while being highly tailored to their preferences.” Not only do AI systems give customers a better service, the ability to scan vast amounts of past and present sales data enables them to pick up new trends and sense the death of ARTIFICIAL INTELLIGENCE HITS THE HIGH STREET Tech journalist CHRIS PARTRIDGE shows how AI is enhancing the consumer shopping experience. Photo:Macys AN AREA WHERE AI IS SHOWING PARTICULAR PROMISE IS HELPING CONSUMERS CHOOSE IN A MARKETPLACE OF A MILLION ALTERNATIVES. STATIONERY BIZ passing fads much more accurately than humans. “The holy grail is great customer service that not only meets, but also predicts their needs and responds accordingly,” de Kerckhove adds. “Only data-driven insights will be able to facilitate this process, with AI and machine-learning moving in tandem with purchasing habits, and instantly identifying – better yet, anticipating – and meeting consumers’ needs.” Many AI systems are now available, such as Microsoft Dynamics’ Einstein and IBM’s Watson, a system that gained worldwide fame when it won the American general knowledge quiz show Jeopardy against human contestants. In that show, Watson demonstrated an uncanny ability to understand casual human chat, including slang. This enables the Watson-powered social media tool Dialogue, developed by digital consultants AKQA, to monitor Twitter and Facebook for statements like ‘I want something like this’ or ‘this is a complete crock’ without naming the brand or company. The system can then post a comment including helpful advice directing the consumer to their product. Crucially, the computer can monitor vast numbers of social media streams for this sort of opportunity, clearly not possible using expensive human beings. Because the system is responding to a cry for help rather than going in cold, the return can be spectacular. The agency claims that 60 per cent of people respond to the offer of help, and almost all of the responses are positive. The online bookseller Alibris uses AI language analysis systems to assess the language skills of its customers, based on their social media posts, and has also used the same tools on the works of more than a thousand authors. Now they can show customers the authors they are likely to appreciate, resulting in people staying on their site longer and a 60 per cent increase in conversions into sales. An area where AI is showing particular promise is helping consumers choose in a marketplace of a million alternatives. AI’s ability to identify things from text, picture or video is a great start. People can make their desires known even if they don’t know the specific name of the item, by simply entering a short description and perhaps adding a picture, and leaving AI to work it out. The system can even cope with conflicting requirements, such as ‘a stylish, slim pen with a big ink reservoir and a low price’, using a technique called trade-off analytics that allocates the weight consumers place on each requirement. Artificial intelligence is not just for online stores, it is also being introduced to help confused customers on the high street. Mobile sales assistants armed with tablets can be supported by AI providing an instant overall view of the stock. This allows them to show details to a customer on any item, on the spot. And of course, almost every customer nowadays has a smartphone that can run a store app. In the US, Macy’s has developed an AI- powered app bringing the information available online to customers in the store, as well as information specific to the store such as the locations of the loos. Crucially, the app can pick up on subtle language cues in the way the customer is interacting with the system, so it knows when to send in a living, breathing human to help. The barriers to adopting AI are lower than you might think. The software is available in the cloud for rent at surprisingly reasonable rates, and hardware prices are coming down, too. In many cases, customers even provide their own hardware in the form of their smartphones running the store app. The biggest danger is public acceptance. To work, an AI system must offer a much better service than a call centre or sales assistants in stores, so it must be designed and tested rigorously before being rolled out to the whole operation. And it must be as easy to talk to as a real human, without pretending to be a real human. The success of Alexa and other digital assistants show most people are perfectly willing to interact with robots, but they need to know who they are talking to, and need to be able to get hold of a real human being when necessary. Augmented Intelligence app in retail Amazon’sAlexa interacts with people just like a human assistant THE HOLY GRAIL IS GREAT CUSTOMER SERVICETHAT NOT ONLY MEETS, BUT ALSO PREDICTS THEIR NEEDS AND RESPONDS ACCORDINGLY. STATIONERY BIZ