If you want to know why businesses are investing in artificial intelligence (AI) technology, ask John Stumpf, CEO of Wells Fargo, one of America’s largest banks. ‘I’m impressed with companies who take complexity and make it simple, who remove pain points… I think those are going to be the kind of things that will be big influences for us,’ he says.
Part of Wells Fargo’s push for simplicity is a smart digital assistant that lets customers ask their mobile banking app questions like ‘how much have I spent on fees?’ or ‘find me my last transaction in this store’ (the app works out which store you’re standing in when you ask).
Intelligence at scale
Most banks expect customers to pick through your bank statements or phone up and ask for help. Wells Fargo is trying to remove that pain point through artificial intelligence.
Retail companies, too, are applying the technology at scale. Paul Clark, CTO of Ocado, is building an automated system for predicting orders and picking and delivering food with as little interaction as possible from humans – including customers. ‘Ocado is a time travel business. We give back our customers the time that they would have spent traipsing around a supermarket,’ says Clarke.
There’s always a competitive advantage in saving customers time and effort. It takes no more than a couple of minutes to Google a taxi company on your smartphone and call them to book a ride – but Uber lets you do the same thing more easily and in just a few seconds.
By successfully removing that small pain point Uber has rapidly grown to be worth $62 billion dollars.
Ready for deployment
The explosion of interest in AI hasn’t come about by chance. It’s a combination of two things: need and capability.
The need arises from our increasing dependence on smartphones. Every sector has seen a huge shift to mobile channels in the past five years. But mobile conversion rates remain stubbornly low compared to desktop. Customers prefer to use their smartphones but it’s just a little harder to get stuff done on a small screen in a distracting environment. AI tech makes it easier and quicker to complete a transaction by cutting down the number and complexity of steps.
With mobile predictive analytics firm Personetics reporting an average 25% response rate for in-context recommendations, it’s clear that smart technology will play a large part in developing next generation mobile experiences.
Meanwhile, the new capability comes both from hardware (today’s smartphones are as powerful as yesterday’s supercomputers) and software (new techniques, such as deep learning, and scalable, robust software that’s fit to deploy at enterprise scale). Along with Wells Fargo, firms such as Ikea, Vodafone, Scandinavian Airlines and Bosch have all been investing in AI technologies.
That combination of acute need and new capability mean that it’s now possible for companies to deploy AI solutions to millions of customers.
The threat of the giants
Among the biggest players in AI are Google, Apple, Amazon, and Facebook. Each of those companies sees an opportunity to create a smart interface between customers and businesses. The smart interface is where the customer relationship takes place and, usually, where the most business value is generated. If companies lose control, they will potentially become nothing more than ‘dumb pipes’.
JP Morgan CEO Jamie Dimon warns ‘They want to eat our lunch. Every one of them is going to try. The businesses that are fighting back are betting that their narrow focus and an injection of technology of their own will help them hold on to their customers.”
Three areas for innovation
cxpartners is helping business leaders to innovate with AI in three ways.
The first is Natural Language Interfaces. Allowing people to ‘talk’ to an app, using voice or text, is often simpler and faster than fighting through menus and tapping at tiny buttons.
Next is sensory input – using your smartphone’s sensors instead of making customers fill in forms. For instance, an app could let you take an insurance inventory of your house from photographs of each of the rooms.
Then comes intelligent agents and deep context. These apps monitor data for the customer and keep an eye out for important events. When something comes up, they can alert the customer or take action on their behalf. Google Now is a good example – it can notice that you have cinema tickets in your Gmail inbox and warn you that traffic is building up and you need to leave early.
Thanks to our partners’ proven technology platforms, it’s no longer a matter of years of R&D to get tangible results – we’re able to get working prototypes up and running remarkably quickly.
Getting it right
Technology by itself is never enough. Worse, a poorly-designed AI can just be a new way to annoy customers with unwanted alerts, confusing them with ill-thought-out responses and damaging businesses’ reputations with uninspiring tech.
Microsoft’s Clippy (the infamous paper clip assistant from earlier versions of Microsoft Office) is a classic example. It was intended to provide smart auto-formatting of documents, but many users were irritated by its ill-judged cartoonish user interface and banal chatter. Microsoft wanted to create a friendly interface, but it ended up patronising and annoying its customers every time they started typing.
Similarly, last year, Yahoo! hoped to improve the findability of photos on Flickr by using an image recognition algorithm to suggest keyword labels for users’ photos. It quickly ran into difficulty when it began to suggest keywords like ‘playground’ and ‘sport’ for photos of World War 2 concentration camps.
What the teams at Microsoft and Yahoo! failed to realise was the importance of the human aspect of design. Algorithms and artificial intelligence need a sense of human etiquette and human sensibility if they’re to be a hit with customers.
Success requires the human touch
Great solutions happen when firms take time to develop a deep understanding of human needs and behaviour.
For instance, when Spotify was building its popular ‘Discover Weekly’ recommendation service, it began by investigating how people listen to music and make recommendations to each other.
The findings contributed to the design of the algorithm that analyses each user’s music taste and makes automated recommendations. The design team also noted that people had clung on to the idea of ‘mix tapes’ even in a digital age. The idea of sending customers a ‘mix tape’ of a couple of dozen tracks once a week proved to be a big hit in user testing.
Matthew Ogle, Product Owner for Discover Weekly at Spotify points out that his team could have launched a service that continually offered thousands of perfectly good recommendations to customers – but this would have felt robotic and inhuman. By building around deep customer insights, Spotify has developed a service that people love.
That’s the biggest lesson that businesses should heed when it comes to artificial intelligence: the secret to success lies in creating services that feel anything but artificial.