When products talk do they have to be friendly?

In Spike Jonze’s 2013 film Her, a heartbroken and moustached Joaquin Phoenix falls deeply in love with his AI-powered personal assistant, Samantha.

And while the plot line might seem a bit ridiculous to most of us, the rise of chat bots has shown just how quickly we’re willing to talk to a machine like it’s a human.

The goal of Samantha, like all technology — especially bots — is to make the user’s life easier. To go from clicking on a website to actually asking your question and getting the answer right away. And while science fiction is still fiction, it does offer a glimpse into how we think the world should work.

Faster is always better, and with the rise in conversational design, natural language processing, and AI-powered bots, talking is edging out clicking as our preferred method for finding information.

At Microsoft’s recent Build developers conference, CEO Satya Nadella proclaimed: “Bots are the new apps”, a statement mirrored by other tech behemoths like Facebook, who has pushed into bot territory with the release of its bots for Messenger Platform.

Even Slack brings bots to your work environment to help with everything from scheduling meetings to finding new and beautiful photography.

For further validation of the current rush in bot development we only need to look across the sea, where the bot uprising has made headway into Asia thanks to platforms like WeChat and Kik, which are used by more than 650 million users a month.

But is this really the future? Will we give up our apps and sites in favor of more chatter?

And if we do, how should it work?

We all want someone to talk to

Chat bots are nothing new.

Back in 1966, an MIT computer scientist named Joseph Weizenbaum published Eliza — a program that would parse words entered into a computer and then match them to a list of scripted responses as a way to mimic human conversation.

Eliza was designed as a mock psychotherapist — something clearly not human. But Weizenbaum was troubled by the response he saw from Eliza’s users.

My actual conversation with Eliza…

When given the opportunity to have a free and open conversation, users started to confide their greatest secrets to Eliza.

Weizenbaum’s reasoning behind building a chat bot was never to create a ‘friend’ or personality that you could engage with, but rather to create a utility. Like an app. In fact, in his book Computer Power and Human Reason, he outright rejected the idea that machines could replace human intellect.

To Weizenbaum, even back in the late-60s, the goal of a chat bot, and all technology, really, was to be just another tool — an extension of the user’s mind.

And, despite science fiction’s utopian vision of a future filled with free-willed AI assistants, most bots these days are just that — a tool. As Jared Newman writes on Fast Company:

“The best way to think about bots, then, is as a new kind of interaction model. Just as mobile apps cut away the sprawling menus of desktop software, chatbots funnel users into a narrow set of interactions that are similar from one bot to the next.”
“Whether that happens through text or by voice, or whether there’s witty banter involved, is secondary.”

Like all good UX design, the idea of a bot is to make this even easier than they already are. To remove friction — the interactions that inhibit people from intuitively and painlessly achieving their goals — and to take away any barrier a user might have from using your product.

And for good reason.

Consider how when Amazon introduced one-click payment, it upped its annual revenue by an estimated $2.4 billion a year.

Or how the Bilingual Child app tripled their revenue when they offered a ‘Buy all’ option instead of making users buy each of their ‘books’ separately.

Bots want to provide these sort of results for every interaction.

Businesses have learned that not everyone wants to download and learn how to use a new app to read the news or order food. Whereas every individual app has its own onboarding process and learning curve, with a chat bot you don’t have to be taught how to use it. If you can type or talk, you can use it.

Does utility need personality?

All this raises an interesting question: If utility and less friction is the goal, do bots need to be friendly?

In a recent post on why bots won’t replace apps, Dan Grover, a product designer at WeChat, examines how Microsoft envisions bots working by looking at one of the most basic (and important) human tasks: ordering a pizza.

In Microsoft’s version, our user chats with ‘Pizzabot’ as if it were a friend. The bot knows his ‘usual’. Knows where he lives. And, assumedly, will charge him directly without him needing to input any payment information.

Utility: Check.

Personality: Maybe a little dry, but sure.

Dan then compares this to his actual experience ordering a pizza from the Pizza Hut bot on WeChat:

The Pizza Hut bot acts less like a friendly helper and more like a guide, giving you only the options you need to get to your goal: ordering a pizza.

In Dan’s opinion, friction can be measured by taps — how many times do you have to click on your screen to get the desired result?

And if that’s the case, then the cold, sterile, Pizza Hut bot wins by a landslide over friendly Microsoft Pizzabot. Less clicks equals less friction. But does it necessarily mean a better interaction?

Dan believes that Microsoft and other bot maker’s ‘conversational’ aspect of design is just another form of skeumorphism — an ornamental design aspect that was necessary in the original form but isn’t any more.

Skeumorphism basically means trying to recreate something analog or from the natural world in a digital space, even if that means including unnecessary designs or features that may have had to occur in the natural world, but don’t in the digital one.

Think about how some of your apps like a calendar or contacts list used to be designed with the same look as their analog counterparts.

The product designers believed we wouldn’t be able to make the mental leap to storing our contacts on our phone or computer unless it looked like what we were already familiar with.

For bots, Dan argues that personality is the same thing — that ‘conversational design’ doesn’t need to include elements of a classic conversation, like niceties, greetings, banter, etc… If those things just produce more friction, why would you choose to use a bot over an app? Once the novelty wears off, it’s the utility that matters.

Just look at Siri. Sure, the virtual assistant knows how to crack a few jokes, but the end goal is always utility.

Ask a question, get an answer. Done.

But, is the future not friendly?

On the other hand, isn’t the promise of a user experience of a Samantha-like personal assistant something we all want? And while bots aren’t being designed as ‘do-all’ AI beings yet, the promise of a more natural interaction is alluring.

Because most bots live in an ecosystem where we interact with other people — like Messenger, Slack, or other chat services — we can’t help but project human values onto them. So why not give them a personality?

But, as Emmet Connolly, director of product design at Intercom, stresses, personality doesn’t mean pretending to be human:

“Don’t pull the rug out from under your users. This means not using ‘is-typing’ indicators or artificial delays to make it seem more human. On the contrary, bot messages should be styled differently and be clearly labeled in a way that communicates they are not human. This doesn’t preclude us from giving the bot personality.”

Personality has its purposes outside of just ‘acting human’.

Beyond just making you feel comfortable and actually want to use the bot, Sebastian Krumhausen, the product and experience designer behind Sure — a Facebook Messenger bot designed to help you find suggestions for sustainable businesses to support — suggests that personality actually helps build a better user experience overall.

“A bot’s personality doesn’t improve the experience in terms of efficiency or ease of use, but it makes it a more pleasant experience. And if you give the bot more personality, users are more forgiving when it fails.”

Rather than be wholly run by scripts, Sure uses what they call ‘human-assisted AI’ where team members can seamlessly jump into a conversation when a question is asked that goes beyond the bot’s current functionality.

So when designing a personality for the bot, they assumedly were designing their own work personalities — friendly, fun, easy-going, relaxed. The kind of qualities that make it easy to forgive if you don’t get the answer you want right away.

“First off, people were very rude in the way they responded back to us, especially when it wouldn’t work,” explains Krumhausen. “But when we added in that quirkiness to the reactions people opened up and became more friendly.”

Marketing scholars call this “brand anthropomorphism” and companies have been using this technique for years.

As Tyler Cowen explains in The New Yorker, decades before bots like Sure gave us quirky personalities to connect with the Dow Chemical Company’s Scrubbing Bubbles told consumers that “we do the work so you don’t have to”; the California Raisins sang “I Heard It Through the Grapevine,”; and the Pillsbury Doughboy giggled about biscuits, rolls, and cookies.

“Corporations create such icons for a simple reason: they know consumers respond to products that seem to engage with them on a personal level.”

With bots, however, the conversation is two-way. Which means that more than just create a personal connection with a brand, a bot has a responsibility to serve, and, as Krumhausen explains, to drive the interaction forward.

“What’s very important is, just like when you’re building a product, it should be super focused in the beginning. So the conversation is really focused. We found it important that the bot sort of nudges or drives the conversation in the specific direction we want it to go in.”

A near-perfect example of this is the Quartz news app.

Just like Sure, the app uses a blend of human-written interactions with scripts. The stories and content are all written by Quartz journalists in the style and tone of their bot and their brand’s voice.

There’s Quartz’s signature brand of quirkiness, with lots of emojis and more than a few jokes and jibs thrown in, but the interaction aspect takes a decidedly choose-your-own adventure route, offering only two options for each story. Read more or move on.

As Jared Newman notes in his Fast Company article, it’s not just the machine side of the interaction that’s scripted anymore. Our responses are as well.

The risk of giving your bot a personality

The Quartz app, while one of the best examples of how a bot could blend utility and personality, also beings up some unique issues.

Just like when you’re forced into a conversation at a party, there’s no guarantee that you’re going to like the person you’re talking to. And even though most bot makers try to create a persona that’s quirky, fun, and easy to get along with, it can get old quickly.

With the Quartz app, their responses at times felt forced, as if distilling an important news story into a version your chill buddy would tell you over beers didn’t really due justice to the event.

Maybe this is just more of what Emmet meant about trying not to ‘fool’ your users.

Personality is fine, but we’ll always know that what we’re talking to isn’t another person. They don’t know us. They can’t read social cues. The goofy use of emojis is just another form of ‘is typing’ humanisms.

Also, as a long-time subscriber to Quartz’s daily newsletter, I did find the app more engaging than scrolling through just another list of articles. But, because it’s a standalone app rather than a bot in Messenger, I found myself wondering if I’d actually go into the app enough to warrant keeping it (storing and opening a standalone app might just be too much friction for me).

And that’s where it fails, as Krumhausen explains in relation to Sure:

“It’s important that it’s efficient and people get that quick win early on. Like any other user experience online, it needs to be fast and efficient. Because if it’s not better or faster than using an existing experience — in our case, something like Google Maps or Yelp — people won’t use it.”

Increasingly, that quick win means being where people already are. The act of downloading the app, and thinking to to open it (rather than read an email in an environment I already frequent) is more friction than it’s worth.

Arguments of personality aside, will bots replace apps?

All of this boils down to the main question around bots: Will they replace apps as we know them?

For most, the answer is no.

Bots, either as a friendly, chatty ‘friend’ or as pure utility are just a part of the tech ecosystem. They offer a chance to reduce friction in key places of your product, whether that’s support, suggestions, or sales.

And in the end, what they do might be more important than what they say.

Ready to build your own bot (before they take over the world)?

Check out Crew and get started working with the best designers and developers on the web. Over 10 million people have used products made on Crew. And over 3 million people have read our blog. Join them here.

Show your support

Clapping shows how much you appreciated Jory MacKay’s story.