A Human Approach to Conversational AI
iGenius Head of Conversational UX, Arianna Stefanoni, explains why empathy is a powerful tool in human-computer interaction
Talking to a machine seemed like something out of a sci-fi movie a few years ago. Now it’s an everyday reality with a name — conversational AI.
This technology is entering every aspect of our lives, with the introduction of apps that assist us with pretty much anything from home chores to answering random questions we might have.
It’s also helping in business contexts, where we can rely on AI more and more to make data-driven decisions, without having to use overly complicated products.
Though, like with any AI concept these days, conversational AI is a massive buzzword out there. So, what is it, anyway?
Essentially, we refer to conversational AI when we talk about applications that leverage artificial intelligence models and strategies to create interfaces that mimic interactions between humans. Or, in simpler words: it’s what enables us to talk with machines.
Beyond Chatbots: Conversational Design
There’s an emerging discipline that conversational AI relies on — conversational design.
Its branches stem quite widely, since the field it operates in is relatively new, and its principles are still evolving. However, we can say that there are three pillars at its foundation:
- A great technical framework — supporting all the engineering complexity behind the user-facing simplicity.
- A multi-disciplinary approach — touching subjects from content design and machine learning, to UX design, information architecture and so on.
- A close attention to human behaviour — being able to observe interactions, both in the real world and in the digital space, and spot useful behavioural patterns that can then be applied to the experience architecture itself.
This last point is crucial, and it is what differentiates conversational AI from chatbots. When users talk with a chatbot, they are 100% sure they are interacting with a machine. On the other hand, when conversational AI is done right, users are left to wonder — “Am I speaking to a machine or to a human?“
To get to this very important point, we’ve had to ask ourselves: “What is “the thing” that makes a conversation human?”
The answer is actually pretty straightforward.
Empathy makes conversation human
Humans have a primal need to be heard and to form emotional connections when engaging in an interaction.
All modern product designers know that the user must be at the very centre of every experience.
Yet one super powerful principle that often gets overlooked is that users are, before anything else, humans.
In one of our research projects, we analyzed conversations users have (without seeing the users’ names) with our AI advisor for data intelligence, crystal.
Of course, we had a lot of business related questions as this is the core of our product. But, we were extremely surprised to find that users also asked quite a lot about what we like to call “human questions”, such as:
- “What’s your favourite pizza?”
- “Where do you live?”
- “What’s your favourite sport?”
- “What’s the weather like?”
Although this might be perceived as the user having a bit of fun, this type of interaction is crucial.
When users experience basic conversational interactions, such as the ones that are generally offered in the current digital landscape, they subconsciously reinforce a pre-existing skeptical belief of “Ok, I am talking to a machine”.
This, in turn, is translated into something like “I’ll only partially trust this because it is behaving in a way that, as a human, I recognize is different from how I behave”.
It’s essential that we let them know that we understand what they mean, whatever that what is, rather than brushing it off with a simple “sorry, i didn’t get what you mean” to gather their trust.
The dark side: dealing with profanity
Having to deal with humans means that we also have to deal with the darker aspects of that spectrum.
This can come in the form of rather impolite user queries — sometimes users really sort of push on the humanity side and spiral into profanity.
The interesting side of this is that this is not done out of, so to speak, bad intentions.
It’s human nature to test the unknown and, in this case, the unknown is the machine itself.
We realized that this, although peculiar and sometimes even quite funny, is an important aspect to cover, too.
Making human-computer interaction simple
As user-centered designers, especially in the conversational design space, we need to make sure we can address all the emotional subsets and engage with it.
Ultimately, our goal at iGenius is to make human-computer interaction as simple as possible while leveraging all the superpowers that applied AI can bring to the table in regards to data analysis!
