Research shows that people respond to technology as if it is another human being. Psychological phenomena like pareidolia, anthropomorphism and the Uncanny Valley demonstrate that we see, perceive and interact with the world through a human lens. It’s because we’re hardwired to do so, thanks to a few hundred thousands of years of evolution.
And that’s why, as conversation designers, it’s our job to teach robots how to speak human.
Language is our superpower
People are experts at conversations. Conversations are part of our lives from a very young age. Most of us don’t really know how it works and why it works, because we understand language intuitively. However, if we want robots to talk like us, we need to understand what it means to talk like us. When language is the interface, we need to understand the intricate ways in which people communicate with each other, verbally and in writing.
A Gricean approach to conversation design
Understanding how we communicate helps us design more human-like, intuitive experiences.
According to the Cooperative Principle, coined by British language philosopher Paul Grice, effective human communication relies on cooperation. And we cooperate in 4 big ways:
- The maxim of quantity: We try not to say too much or too little
- The maxim of quality: We try to be truthful about the things we say
- The maxim of relation: We try to be relevant to the conversation we’re having
- The maxim of manner: We try to communicate as clearly as we can
A conversation is teamwork
Knowing people rely on cooperation, you will need to design accordingly. For starters, it means we can’t overwhelm people with large chunks of information without breaking it down for them to understand. That’s not helpful and we’re not adhering to the maxim of quantity.
It also means we should expect users to be cooperative. And because users are, we have to assume they occasionally offer more information than is literally asked of them. In turn, your AI Assistant should try to actively move the conversation forward and be relevant and useful throughout. In doing so, we’re adhering to the maxim of relation.
Surely, your AI Assistant won’t always be able to handle every single response. In cases like these, you want to rely on elegant repair messages to get the dialog back on track. Of course, always in a way that doesn’t assign blame on the user.
The importance of turn-taking
Another important concept is turn-taking. It’s a way to describe the back-and-forth nature of communication. The word is pretty much self-explanatory: people take turns when speaking to each other.
Understanding the concept of turn-taking is crucial for a conversation designer. Normally, people give each other all sorts of cues to let each other know when it’s their turn to speak: turning to someone, gesturing, a nod. However, with chatbots and voice assistants we have limited real estate and the focus lies more on the language.
Ending the AI Assistant’s responses with a clear prompt is a very effective strategy. We can also guide the user’s focus through word order, stress, and vocal pitch. And by actively moving the conversation forward, keeping responses short and sweet, and asking follow up questions your AI Assistant can become a true conversational partner.
Designing better experiences starts with fundamentals
Knowing the ‘rules of conversation’ will help you design better conversations. We want to design conversations that match our human conversational strategies. It’s a prerequisite for creating successful AI Assistants and one of the most important drivers for adoption. Only if we do that, will we succeed in creating trustworthy, reliable, and intuitive experiences.