Tom Greenwood, senior designer at Designit London, advises CEOs to define the strategy that makes their company more human-centric.

 

The media is touting 2018 as the year that artificial intelligence finally comes into its own.

 

Companies are finally taking AI seriously as a result. But here’s the rub: Clients drink the AI-is-coming-for-you Kool-Aid and want to move fast — faster than the capabilities of technologists and designers to help them. Artificial intelligence isn’t an add-on, it’s something you build into things.

 

Designit: What are the biggest mistakes clients are making?

 

TG: The ethos for AI needs to be human-centered, not tech-centered. Companies tend to invest in a technology or develop a product, and then wonder how to apply it. Our goal is to get our clients to flip that technology-led process on its head. Ideally you do research and find the real human needs first, and then you develop the technology. There’s a lot of working through possibilities and running models to see what can work. What are the repercussions if you apply a technology this way or that way, and does that get you what you need from the customers you’re after or the product you’re developing? Does it create new problems? Does AI make your company more or less human-centric? If your focus is people, not revenue, later you’ll reap the financial return.

 

Designit: What are the most promising opportunities in AI?

 

TG: Customer service in banking. Traditional banks have a problem: They need to differentiate themselves from startups. The natural way to do that would be to improve customer service. This is a longstanding issue that’s deepening now that people have 24/7 access and countless touchpoints. People are impatient. They expect fast results, immediate responses. Do you remember the last time you phoned a bank? You probably got stuck listening to hold music.

 

Designit: So what’s the solution?

 

TG: Use AI to give the customer a more empathetic experience. Put a service design lens to the issue and it quickly becomes obvious: Can AI help a customer service rep do his or her job? What about that person’s colleagues? In what ways, exactly, can it help businesses improve their interactions with customers? Do we use AI to automate certain processes in the loop or to help the human?

 

Designit: What purpose do chatbots serve?

 

TG: Chatbots are good at solving simple and common queries, but if all they do is replace the human interaction, that’s no solution. People want to talk to a human, and they want those reps to be empathetic. That’s why AI works best when it plays a role in customer service. What it’s really good at is non-human tasks, such as collecting data about a person and connecting the dots. A well-designed chatbot knows when to hand off that data with helpful suggestions to support customer service reps — and when to just stop the conversation and bring a human into the loop.

 

Designit: What about the ethics of using data and AI in business?

 

TG: This is going to be a big deal in the next few years. How comfortable are people about letting their data be collected, and who wouldn’t they mind sharing that information with inside their organization? From the product side, what benefit does a product have to offer in exchange for someone to surrender his or her private data? I spend a lot of time looking at the ethics of collecting and using data while providing the best possible user experience. It’s not always an easy or comfortable mix. Say you’re collecting someone’s personal data about his workplace productivity. Ideally, the data provides meaningful insights for him, but also for the organization. You want transparency of data, but at a level customers find acceptable.

 

Designit: Is AI coming to take our jobs?

 

TG: It’ll take some, change others, and create some new ones. What’s more compelling is whether CEOs truly understand how AI is affecting them and whether they have a strategy for the future. Because artificial intelligence willtransform business in two key areas: customer experience and automation. On the front end, voice recognition, chatbots, and biometrics are enhancing the interface. On the back end, processes are becoming more efficient through predictive analytics and process automation (for example, automating loan applications across departments). How many companies have a strategy that combines both elements while also adopting a change management mindset within the organization?

 

Designit: What attracted you to this field?

 

TG: While studying for a masters in human-computer interaction [HCI] I joined a research program for an Italian marine engineering company that was exploring how autonomous underwater vehicles could be controlled by divers using gesture control. Diving’s my passion, so I jumped at the opportunity, figuring I’d get to go dive in the Mediterranean. I ended up in London in a pool. Still, I loved figuring out how people can communicate with machines in a way that’s natural. I developed a language of gestures for divers using signals that they already used, and then studied where it would break down and how to define those signals for a robot. Having the HCI background really affected my work in AI as a designer.

 

Designit: Did this help you understand AI’s true potential?

 

TG: The underwater autonomous vehicles could do things humans couldn’t — dive deeper, pick up heavier objects, stay submerged for longer. It helped me think of AI as an assistant or a co-worker with different skills, not as a machine that will replace us. Right now, for example, machines help doctors diagnose diseases, but they lack the emotional intelligence for a good bedside manner. At least for now. Designers should be right out there in the forefront, helping to define AI’s supporting role.

 

Designit: What needs to happen next with AI?

 

TG: Two things are missing right now. First, a lot of the solutions being rolled out aren’t well designed. The experience is terrible partly because the technology isn’t quite there. The second problem is AI is not yet being designed from a user-centric point-of-view. Here’s the standard MO: “Hey, we’ve got that chatbot technology, let’s use it, it’ll save us money.” It’s rolled out for wrong reason.

 

Change is underfoot, it’s just not fast-moving. Now we’re starting to see designers working closely with data scientists, and that’s exactly what needs to happen. With hackathons now all the rage there’s generally an eye to taking a more collaborative approach. But designers can’t design AI without consulting experts in writing algorithms. It’s equally true that AI engineers can’t work without designers who understand usability, ethics, and emotional design. It’s coming together, but mainly in blockbuster companies such as Google, Apple, and Amazon, which suck up the top talent from machine learning programs at MIT, Harvard, and Cambridge.

 

Multinationals like us just follow the talent wherever we find it: We have a cognitive computing team in Bangalore that’s helping us develop our collaborative approach to AI.

 

Designit: What about everyone else?

 

TG: Everywhere else is where the nitty gritty customer service work is being right now. A bank, for example, doesn’t need superintelligence, they just need to take a holistic service design approach when designing for customer service chatbots. Those chatbots should have the right personality for the job. They should also have the ability to gather data in ways that help them make predictions or suggestions that personalize the customer experience. And then the bank needs to roll it all out in an efficient way.

 

Designit: How do you design a brand’s chatbot personality?

 

TG: It should suit the brand and use case. You use the same technology on the backend so you don’t have to start from scratch with each chatbot, for different products. Then you tweak that, design this, and build one chatbot for mortgages and a different one for payments. One for retired folks and another for young people. Each of those chatbots may need to have a different tone of voice. Designing a personality is a little more difficult. Part of it is timing: ensuring that the chatbot knows when to bring a human into the conversation, and when to bring in the data it gathers. On the simplest level, it happens to be the customer’s birthday, so the chatbot nudges the human, Hey, say happy birthday! That’s where the design bit comes in, understanding data points and adding meaning to them.

 

Designit: What happens when people get angry at chatbots or they just show up angry?

 

TG: Let me sidetrack before I answer. People are actually more open when talking to artificially intelligent machines. They find it easier to talk to a machine that isn’t judging them. So if you have an AI-based therapist using voice-enabled AI, it changes people’s behavior.

 

That said, therapy is different than problem-solving your mobile account.

 

One answer is this is where personality comes in. The more human you make an AI machine or product, the more natural the interaction becomes. The flip side is that the expectations of the intelligence also increases.

 

What happens now is that something appears to be having a conversation with you, but the technology behind it is actually way behind. If we build empathy into a chatbot however, it can change the way it responds to you, by recognizing emotions in your response and changing its behavior. Maybe it uses humor or softens its tone, or alternatively it feels like switching you to a human immediately, and offers that as an option. The important part in training the AI in the right behavior.

 

 

Designit: What kind of AI projects do you have in the pipeline?

 

TG: Our new London office has a Maker/Venture lab with our tech partners Buildit, and we’re experimenting with our first product, an AI-powered workplace productivity tool called Totem. It’s a six-sided digitally connected cube that sits on your desk. It’s sort of like a productivity sidekick. It collects your data during working hours. It follows your work style and updates your work mode accordingly. It can change your status on Slack, mute your notifications, suggest a playlist on Spotify. Then it finds patterns, shows you your patterns, and asks you how productive you feel, and whether that time is meaningful for you. Eventually it starts comparing your working patterns and productivity.

 

Designit: What’s in it for the user? It’s a little Big Brother-y.

 

TG: More like the quantified self for work. It’s got a micro-trend value for the user, because it helps people do more meaningful work. For the organization, it looks for macrotrends to help the business get more productive. Companies have never had a way to capture and understand these kinds of patterns, so it opens up all kinds of possibilities for smarter workplaces.

 

Designit: How far into the design pipeline are you?

 

TG: Less than a year out from beta. We’ve designed three prototypes that we’re testing daily. The product works by showing you your work modes daily, giving you a summary at the end of the week. Over time you work out your patterns and Totem makes suggestions based on those patterns. Maybe you’ll learn that you do more meaningful and deep conceptual work in the morning and more tangible work in the afternoon. Perhaps even small changes to your working pattern will reshape how you feel throughout the day. Once you’re incentivized, you’re more engaged. Eventually we’ll refine the concept of how Totem will work on a social or organizational level — maybe a social function kicks in and everyone switches to pause so you can all go out for coffee together.

Tom Greenwood

Tom Greenwood

Senior Designer, Designit

@tomgreenwood

Tom Greenwood is a Senior UX Designer at Designit, a Wipro company. Designit works with ambitious brands to create high-impact products, services, systems and spaces – that people love. Because what matters to people, matters to business.

What you’ve read here? Tip of the iceberg. Are you ready to be part of the excitement?