The first rule of chatbots: Don’t lie to your users An interview with UX designer Joe Toscano

Chatbot

By Vishal Gangawane

Chatbots were all the rage in 2016 and 2017 promises to be no different. We spoke to Joe Toscano, an user experience design professional based in San Fransisco, about the current and future impact of chatbots. Toscano manages and lead experience design and prototyping for R/GA’s team embedded at Google. R/GA is a full service digital agency creating products, services and communications to help grow our client’s businesses in the connected age. He also blogs on the Invision blog and at Muzli.
Q. As a primer, could you tell us in brief about chatbots? Also, are chatbots truly the next big thing and if true what industries is this likely to disrupt?
Chatbots, for those that are unaware, are software programs created to replicate human conversation with human users.
The concept of chatbots has been around for several decades. Since the 1950s, to be exact. The beginnings can be traced back to Alan Turing, and his experiments with computer science and human intelligence. These would later form the basis for what we refer to as the Turing test.
Creating a true “chat” bot is very difficult. It requires a lot of data and a lot of iteration.
I believe chatbots as most people understand them are a trend at this moment in history. I think right now we’re going to see a lot of companies trying to own the space and create the best bot possible. And I think for that reason chatbots are going to disrupt a lot of industries.
But I think once our natural language processing systems have enough data running through them and we can speak to these bots, the bubble will pop. I think the modern chatbot manifested in a text messaging app will be around for a while, maybe 3-5 years, but I don’t think it’s the end all product that most bot makers and creative technologists are truly excited about. I think chatbots as we currently understand them are just the onboarding for the future.
Q. You have written about your best practices for building chatbots. Could you repeat them for our readers again in brief here?
1.Don’t lie to your users: I think this is important for any brand/product. But it’s especially important with chatbots. Humans are much less likely to trust a computer agent than another human. And if you blow it once, you’re probably not going to get a second or third chance.
2. Onboard with conversation: The general public isn’t used to interacting with and commanding their devices. Help them learn by initiating the conversation. Don’t create a tutorial, but ask them questions or give them commands about how to use your bot.
3. Design for human emotion: Similar to point 1, your bot is not inherently attractive to human users. We, as humans, crave human connection. It’s important that your bot meets your user as a human. But don’t try to trick your users into believing it is a human. It’s a fine line.
4. Conversation is limitless: Language is an incredible tool. It’s the closest tool we have to mind alteration. We can speak and communicate ideas from one head to the other with relatively simple ease. But teaching a computer to understand language like we do is not easy.
5. Create boundaries: We’re better off creating specific conversations and directing the flow of the conversation to keep people within the ‘loop’ we’ve created. It will not only help make sure your bot doesn’t break, but also make sure that your users are having the best experience possible.
6. Let them down easy: When your bot does fail, make sure you’ve created a plan to help keep people around. If your screen goes blank or you give them some painful error message, your users are just going to leave and probably not come back again.
7. Every interaction is meaningful: Unlike the web and apps where many interactions are required just to navigate the page, every interaction with your bot will give an output. Every interaction becomes meaningful.
8. Help users help you: You’ll never know everything your users want. That goes for any product. But bots make it easier to figure it out. All you have to do is ask. Or offer a spot for your users to submit things. Let your users shape the product by telling you what they want.
9. Identify and target user sentiment: As we’ve already recognized, human emotion is as important in bots as anywhere else on the web. Maybe more important. And it’s the first platform where we’re getting insights into human emotions through contextual conversation. We can use this to teach our bot to recognize the emotions and help create a better experience.
Q. Say five to ten years down the line, how do you envision chatbots transforming the lives of users?
Humans are very habitual creatures. Sure, not every day is exactly the same, but there are a lot of similarities across days. I believe the knowledge we get from bots will allow us to build systems that anticipate our needs and get us away from our screens. I believe eventually bots will become an extension of our self.
Right now we do this through sites and apps that automate things for us — Reminder apps to keep us on track, apps like Instacart to do our grocery shopping, apps like Uber to get us a ride from place to place.
I believe eventually we won’t have to pull our phones out of our pocket but we’ll be able to make all these things happen.
Q. In the same vein, how will the role of UX designers change in about the same period?
I believe the role of UX designer will stay relatively the same — create systems that meet your users needs in the most intuitive way. But I believe the needs of users will change, and that’s the difference.
That’s just a historical fact though. If we stayed the same, society would never move forward. I just think we’re going to start moving forward at exponential rates. I’d argue we already have in many parts of the world, but I believe it’s going to be global instead of just in pockets of the world like San Francisco, New York, South Korea, etc.
Q. Aren’t chatbots essentially a technical project? Will UX designers have a role to play beyond providing the script/dialogue of how the conversations will play out?
Similar to the last question, I believe UX designers will always have a role — understanding and empathizing with the humans using the product. I believe this will be one of the last roles to be phased out of technical projects, because human emotion is so hard to replicate in binaries.
Q. How will chatbots affect enterprise software? India’s software industry primarily caters to large enterprise clients overseas so we are extremely interested in seeing how this plays out.
I don’t have specific examples off the top of my head but any way that systems can be automated and made more efficient, an enterprise is going to love it. And I know that’s a fact.
Q. Will the emergence of chatbots mean lesser investments in building web, mobile applications by enterprises? Or will chatbots play a more complimentary role with what gets built?
I think initially you’re going to see businesses trying to build chatbots within their apps. I think it’s going to be difficult for businesses to commit to the fact that apps are going to be old news, because they’ve invested so much into their ecosystems and it won’t be easy to just toss that stuff out the window. I think that’s part of why it will take so long for bots to become the default.
But I believe eventually bots (i.e. Google Assistant, Siri, Alexa, etc.) will become the next telecom. I believe there will be a few that hold the market and people will use them based on what their algorithms produce.
Data is the new oil and building the best AI system is the new race to space.
Q. Which companies/startups/people are the ones to watch out in the chatbots arena?
In terms of complex, data driven bots used for search and information/navigating the world around us:
• Google → Google Assistant
• Amazon → Alexa
• Apple → Siri
• SoundHound → Hound
In terms of focused bots performing specific tasks that do it really well right now:
• Pana
• Hipmunk
• Duolingo
• Penny
Q. What are likely to be the hurdles and roadblocks before we see mainstream adoption of chatbots?
I believe the biggest hurdle will be user adoption and I believe the biggest roadblock to that will be making sure bots do not turn into spam messaging bots, because bots already have a stigma of being spam and if we want people to adopt these systems, we’re going to need to turn that around.
Also, privacy. Making sure people know what their data is being used for and how they can control where their data is going.
This is a very important chapter in my book, actually.
Q. And finally, is the hype true? Will chatbots kill apps?
Yes, I believe the hype is true. I think it’s still going to take 2-3 years before we have a revolutionary breakthrough that makes these technologies accessible to everyone, and I believe it will take 3-5 before bots become adopted by the mainstream, but I believe automated systems that are backed by artificial intelligence and are interacted with through voice/text will kill off apps eventually.
If you want to know more about chatbots, Toscano in the process of writing a book that explains all these points and much more. It is likely to be done by mid-February 2017. You can sign up to join over 700 people that are already on the waitlist here: http://bit.ly/2eELh9W
Recent Events

Recent Events

No posts available

 
Clip to Evernote