Designing Conversations — Giles Colborne
Notes from UX Salon 2017 — by Summurai
Are conversational interfaces the next generation for user interfaces? This presentation examines conversational user interfaces and why they offer a new opportunity for faster more efficient interaction. It also examines the pitfalls of such user interfaces and the psychology of human conversation.
A bad, non-intuitive UI design easily frustrates users. Take for example the challenge of fitting the London train times schedule into a small screen. Not only would this frustrate the user, it would also discourage them from using this interface to get the information they need. This in mind, using a chat interface instead, one where the user can send a message asking what time their train is leaving and receive a response with the exact information they need in order to catch their train in the most efficient way possible. The future is leading us to a natural conversation with virtual assistant bots. Bots, as we know them today are mostly terrible to interact with. The UI is click-oriented and usually includes filling out forms. Siri allows you to have a hands-free conversation with your virtual assistant which is great in private spaces such as your home, but not so great in public spaces like your office or the train station.
Bots are usually designed with decision-trees and lack the natural language elements like being able to change one’s mind as the conversation continues, about a question that was previously asked. Humans in this case, will provide the most relevant short version of data. Questions such as “Where is the train station?” Should be answered with easy answers such as “straight ahead instead of “300 meters north…” Another way to make voice interactions more user friendly is by using a personality for the bot. Giles gives the example of Grunk, a cartoon pig with a funny accent. When Grunk doesn’t understand you, it’s more forgivable than when Siri doesn’t understand you, because he’s just a pig.
Trying to create a natural and friendly conversation with bots such as siri makes the expectations from the other side much higher. Another example is the use of emojies. The most used emojies are the ones that convey emotions. Bots should be able to use emojies, the same way humans do. This illustrates design with social and emotional sensitivity.
The facebook year in review feature which presented users with a summary of their year can be especially insensitive in cases where the user actually had a horrible year, such as the death of a family member or dear friend. This is called “Algorithm’s cruelty”, and stresses the importance of considering various social situations and contexts within the design process. A good bot should listen and empathize. Lastly, it’s important to remember that we are not designing friends. We are designing assistants.
Giles Colborn is a co-founder of CXpartners — one of the world’s leading independent experience design consultancies. He is also the Author of. ‘Simple and Usable’.