The widespread adoption of conversational interfaces replaces the master-to-slave narrative (e.g. artist using a paintbrush) with one that’s more human-to-human. Bots are crucial to these developments, which are gaining momentum across popular social platforms (Slack, Facebook messenger, kik and many others) as well as in other unexpected places. We’ll explore some general ideas on how to keep interactions with bots meaningful and enjoyable for everyone.
Computer Literacy vs Digital Natives
Texting someone on Whatsapp, using Slack to talk to coworkers, or speaking to Siri for a pub debate fact-check; these are all examples of conversational interfaces. Even the green-screened command-line used in countless Hollywood hacker tropes is a conversational interface. It’s not a new concept, instances of them are becoming more prevalent than ever before due to shifting online demographics. People are more digitally literate now than 5 or 10 years ago. We’ve gradually stopped talking about “computer literacy” as a skill that’s distinct from being competent as an average person.
Digital experiences are becoming more conversational because people naturally engage each other in conversation to get things done. Many companies are realizing how useful this option is to building experiences that are inherently social in some way. The primary advantage offered by conversational interfaces is the universal reduction of mental effort needed to design, develop, and comprehend a user journey. A quick check: what feels more human to you? Saving or “Hearting”? Up-voting or “Reacting”?
Coherence, elegance, and fluidity are the end result of a meaningful social exchange. Using this UX paradigm carefully can alleviate interaction designers from working to build an arbitrary context. Instead they could rely on human social protocol to support the user flow. Doing so would free up those skilled resources to be focused more nuanced aspects of a digital experience.
A New Way of Thinking
Active collaboration with third-party writers, designers, and developers have allowed bots, which operate as human avatars, to expand platform capabilities. Fundamentally, this is the same as what mobile app stores do for smartphones.
The vast majority of discussions on bot technology have been focused around tools that help developers build bots. Out of the oceans of articles written on the topic of bots inside conversational interfaces, few touch on the design impact of bot personalities. Clearly, studying the social protocols of bot conduct is a side of the subject that is understated in importance and underserved. Bots have some distinct advantages versus apps in this regard:
- (A) Easier to understand
Bots can communicate with each other as well as humans using natural language instead of a programmatic language. - (B) Well-defined habitat
Encounters with bots are generally initiated and concluded inside the messaging platform. A bot does not leak its presence externally. - (C) Singular purpose
Each bot can be expected to do exactly one thing: a minimum useful subset of a human role within its environment.
Six Bare-bone Behaviors
It’s not hard to come with an overwhelming list of things to think about in devising ways to make bot interaction better. Perhaps a better approach would be to ask “What’s the minimum needed in creating bots that are socialized and maintain their natural advantages over apps?”
- Bots should be named and respond to their names
Just like humans, depending on their role they should be named aptly for the tasks they can help you with. A bot should always reply to a mention of its name for continuity within the conversation. - It should be easy to figure out how to trigger the bot
As a possible reaction to a mention of its name, a bot should tell the conversation participants how to get it to do things, either by a specific example in the immediate reply or suggesting a “help” trigger word which it can then use to list its functions via PM (private message). - Bots should be verbose
If the purpose of a bot is saving humans time and energy, it’s best to keep humans up to date about what it’s doing so its actions are transparent. It might also motivate people to give more useful bot improvement suggestions. - Avoid the uncanny valley when writing what bots say
This fits with the advantage of being easier to understand. Evidence suggests that people perceive bots favourably and react to them productively when bots don’t try too hard to be human when the illusion isn’t strictly necessary. Though bot phrases should be well formed sentences, there’s a natural allowance within dialogue as long as the audience understands its language. - Any bot should be able to trigger other bots within a conversation
By writing these triggers and reactions in plain English (or other native language), a single bot can expose how to get other bots to react, and do things beyond its immediate responsibility. The side effect is that conversation between bots socially reinforces the concept that bots have human-like agency. - Bots should defer to humans more as tasks get more complex
Tackling automation of low hanging fruit is one way to keep our narrative simple. If a particular task requires high cognitive load to understand or complete correctly, it’s a task that might be better being totally offloaded to humans. This is true especially if it takes a lot of time to correct the bot to interpret our intentions in the way that we want or fix its mistakes.
A comprehensive resource of social patterns and behaviors would be essential as more designers and developers delve into the art of crafting bot experiences. Perhaps we might go about this by adapting the list above as a starting point to formulate these patterns. Has this been useful to you? Add your comments, suggestions, and feedback below.
Stay tuned for more upcoming bot posts from the This Place team!
