Natural interaction thanks to Conversational user interfaces (CUI)
They tell us what the weather will be like. Play our playlist on command. Or remind us of appointments. Be it Cortana, Siri, Alexa, Bixby or Google Assistant – conversational user interfaces (CUIs) have long become an integral part of our smartphones and of our sitting rooms. And what is more, conversational user interfaces are not limited to voice assistants – chatbots or voice user interfaces also fall into this category. This so-called natural language interaction is often celebrated to be a paradigm shift and the future of interaction. After all, it has always had a reputation for being the ultimate intuitive interface. Totally effortless and natural. Up to now, this promise has not been delivered on. Will this change now thanks to conversational user interfaces?
A journey through history
Conversational user interfaces are nothing new. In fact, the interaction with computers started with speech. People used written language to give commands, and computers executed these instructions. The disadvantage: The users had to remember the commands with their exact syntax. Therefore, the first graphical user interfaces (GUIs) were a real revolution. They displayed digital objects metaphorically as files and folders on a desktop. Instead of speech commands, there were "physical actions" like opening, moving or deleting documents.The interaction using natural user interfaces imitated reality even more. Reality-based interaction incorporates various characteristics that users have become acquainted with in their everyday environment since their childhood. For example, it simulates physics: You can start or stop scrolling through lists on a smartphone at the touch of a finger. This emulates the effects of momentum conservation and friction that the users are familiar with from their everyday lives. And even though speech commands played a very limited role, they are now increasingly popular due to technological progress: Thanks to natural-language procession (NLP) and machine learning (ML) users can nowadays talk to a computer quite naturally.
Talk to me – CUIs in use
It is therefore not surprising that our lives these days are hard to imagine without conversational user interfaces. They are especially useful for sales and front offices: Conversations with customers are, in a way, outsourced to machines. In this context, a distinction is made between chatbots and voice assistants. Chatbots are something like a text version of conversational user interfaces. They imitate an authentic conversation based on written texts, e.g. in customer service.
Voice assistants, however, simulate a natural conversation: Users can actually talk to and, to a certain extent, with them. The name already makes it clear: They should assist the users and make their lives easier, from online shopping to answering questions, searching the web and controlling the smart home. And the smartphone is no longer the only device to use since voice assistants have moved into the homes in the form of smart speakers such as Apple Homepod and Amazon Echo. In 2019, 32 % of Germans used a voice assistant. 47 % used their smart speaker on a daily basis. However, research has shown that the activities are mostly limited to rather simple actions – the much-heralded revolution in human-computer interaction has yet to materialize.
The benefits of CUIs
People can use their everyday language. This makes the familiarization process easier and the usage more intuitive. Moreover, users can interact with the computer even when their hands aren’t free. The same is true when they have to visually focus on other activities such as driving. Therefore, conversational user interfaces are easily accessible even for people with motor or visual impairments. And the CUIs even adapt to the speakers, their intonation, speech volume or rate of speech.
Are appearances deceiving?
Despite all the benefits, there are still some problems for which conversational user interfaces have no solution yet: Oftentimes, users have no overview of the possibilities and limitations of the system. They need to try them out – which can be a tiring and frustrating process. Users may get the impression that the system has comprehensive functionality and that they can actually talk to the system. In general though, the CUI systems turn into mere search engines and give up.
Siri, we didn’t mean you
Conversational user interfaces that are controlled via voice commands always have to be addressed, be it through a so-called wake word such as "Hey, Siri" or another form of interaction such as pushing a button. However, the system’s turn-taking capabilities are still imperfect. This also means that Alexa, Siri & co. erroneously feel that they have been addressed and deduct activities from the conversation that the users haven’t intended.
Who are you talking to?
The presence of other people, too, often leads to misunderstandings: If a person starts a dialogue with a conversational user interface, other people in the room might think that they are being addressed, wrongly interpreting the dialogue and reacting accordingly. This is not only unpleasant for the person who hasn‘t been addressed but also for the CUI user. The consequence: CUI users reduce the speech volume, which in turn means that the conversational user interface doesn’t understand the command and asks the user to repeat it. It is therefore often difficult to use a conversational user interface in the presence of other people.
Ethical questions, too, have severe impacts: What data does the conversational user interface collect? How are they used? For the user, this is often impossible to find out. Another problem: The CUI Google Duplex can be used to make appointments. With Google Duplex, this activity has become so natural that the users don’t know whether they are talking to a real person or to a technology. From an ethical point of view, it would therefore be important that 1. it is obvious for users whether they are talking to technologies, 2. data are treated confidentially, 3. users have the opportunity to consent to a conversation.
Conversational user interfaces can make interaction more natural and open up new forms of interaction. In future, they may unlock their full potential not only in the smart home but also in business contexts, especially when humans and artificial intelligence cooperate. In general, however, designers and developers have to decide in what cases the use of conversational user interfaces is appropriate. To this end, the chances and challenges detailed above will hopefully be of assistance.
Prof. Dr. Michael Burmester is UID’s Principal Scientific Advisor. Since 2002, Dr. Michael Burmester has been holding a chair as professor for ergonomics and usability in the information design course at the HdM Media University in Stuttgart. His research work deals with methods of human-centered design and positive user experience. The main focus of his current research projects is positive user experience in modern working environments including current technological developments such as artificial intelligence, virtual reality, augmented reality, etc. At the HdM he heads the User Experience Research Labs (UXL) and the Information Experience and Design Research Group (IXD). He also acts as a coordinator of a project financed by the German Ministry of Economics that supports SMEs in designing digital technologies with positive usability and user experience (Mittelstand 4.0 – Kompetenzzentrum Usability).
- AI in focus: Positive experiences in human-AI interaction
- AI in focus: Artificial intelligence & positive UX
- AI in focus: Human-robot collaboration
- AI in focus: The cards are being reshuffled
- UX Buzzword Jungle: Better products thanks to Lean UX
- UX Buzzword Jungle: Augmented Reality: between fascination and challenges
- UX Buzzword Jungle: Home, smart home – theory and practice of the connected home
- UX Buzzword Jungle: Intuitive operation – a positive user experience without instructions
- UX Buzzword Jungle: Gamification – a positive way to playful user experiences?
- UX Buzzword Jungle: User experience = Usability plus X?
- UX Buzzword Jungle: Design Thinking – new old creativity