This is Part 1 of my series on How AI would impact UX and vice versa. This write up covers different facets of chat bots and the resulting experiences.
Hari Nallan & Symran Bhue - July 2017
In an Artificially Intelligent world, pairing future-tech and design will result in incredibly innovative products which will transform our lives. But for that, the User Experience and Interaction Design practices will need to be different from traditional design practices.
Traditionally, a design team would research users, study their behavior, expectations and goals in certain scenarios and during the research phase – patterns related to the users emerge. These patterns when broken down – would be translated into Personas. Based on different courses of action a Persona would take in different Scenarios, the design team would define User Journeys and so on.
With more and more bots representing organizations and products, it is here, where organizations must re-evaluate the Interaction design and User Experience that they are providing the customers. In this context, thinking of a chatbot only as a UI element will qualify as poor design practice.
Bots with a Persona:
Conversations come naturally to humans, but with recent developments in the technology, the idea of conversations between humans and the artificial intelligence is not as alien as it was a decade ago. The recent successes in the area of AI has been seen in language translation and questions answered by chatbots; However, whether the User Experience of this interaction is pleasant or frustrating depends on user expectations and the purpose of the virtual assistant. For instance, if users are expecting intelligent assistants to have an answer to their questions, then whether or not the bot has empathy, may not be of the topmost concern for such users. On the other hand, if users are expecting a good customer service bot, then the level of friendliness of the virtual agent would either please or disappoint the users.
PWC sampled 2,500 US consumers and business decision makers via a nationally representative online survey to explore attitudes towards Artificial Intelligence and its current and future implications on society. 27% of the sampled consumers said that they weren’t certain whether their interaction with a business or brand was through a human or a virtual assistant.
This can only be accomplished if a intelligent agent is rendered a persona. Now whether the persona is that of an existing human customer service agent or an ideal but imaginary customer service agent is a decision businesses need to make. But if the persona is that of an existing customer service agent then the agent could drop off or come on as he chooses because while the agent is away, the virtual agent could take over and customer’s queries could be answered without interruptions as well as seamlessly.
Scenario Stories with surprises:
When users decide to chat with a bot, or a human from customer service – it is usually because they need assistance with a scenario or a query they may have. The interesting part about this is, in case of a chatbot, it could make recommendations to the user straight away instead of asking them a question.
For example, a collaboration platform such as Bitrix24 is available for desktop as well as mobile but the mobile version has certain limitations such as the ability to edit one’s profile. Users accessing the platform asking for support to edit their profile through the mobile app could be straightaway recommended by Mia (the friendly Bitrix24 AI Assistant) to switch to desktop version of the app. The user wouldn’t shy away from thanking her for such pleasant surprises.
The point is that the virtual agent could know user preferences, mode of access and other user specific data to make recommendations on the next steps or suggest user journeys which wouldn’t be a standard course of action for the user but could be the most efficient journey for that user.
User Journeys, initiated or uninitiated:
User journeys would include not only those elements initiated by the user but also elements whose initiation would be influenced by the “intelligence”. User actions are part of user journeys which would change based on whether the user took an action without any intervention from the bot; the user acted as per the artificial agent’s recommendations; the user ignored the artificial agent’s recommendation in taking an action or the virtual agent acted on behalf of the user.
Poncho, the top ranked chatbot on ChatBottle, is a Facebook Messenger weathercat from Brooklyn. When you chat with Poncho, he starts off by introducing himself and then asking for the user’s location to forecast the weather. Poncho follows a standard script of questions and the courses of action a user can take are also defined clearly to keep the user journeys predictable
However, the interaction with Poncho is engaging because he couples weather forecasts with funny gifs to keep the user entertained.
One much needed enhancement in the current user experience with the AI assistance would be designing for context. Most bots are designed to respond to one question at a time without the context of the current conversation at hand. This implies that the bot will have the same answer for the same question asked multiple times in the same conversation irrespective of the context. As soon as you enter your location for Poncho to forecast the weather, Poncho’s immediate response is that – his ex is from that location; However, if you ask where his ex is from – he doesn’t have the context to answer to that question even though he brought up the subject.
What have your experiences with chatbots been like? Have you interacted with an ‘artificial intelligence’ that was designed to remember context? Tell us about it?
Watch out for my second blog on this theme where we discuss the impact of AI on mental models, user goals and KPIs. View Part 2
