What are chatbots and voice assistants really capable of?
In 2018, Google gave us a good impression of how digital communication assistants would be able to support us in everyday life. Unfortunately, our current experience with chatbots, voice assistants, and the like leaves a lot to be desired. Repetitive, monosyllabic answers in chat or a “Sorry, I can’t help you there” from Siri or Alexa – who doesn’t know these scenarios?
In order to be able to install better speech and text-based assistance systems, some assumptions and prerequisites are needed. We also need to be aware of the natural limits of these systems. And these are precisely the topics and possible opportunities for digital assistants that we cover in this expert view.
What are digital communication assistants?
When we speak of digital communication assistants, we mean both text and speech-based systems such as:
- chatbots on websites or in messengers (e.g. Slack and WhatsApp)
- voice interfaces or voice assistants (e.g. Alexa and Siri) as well as assistance systems that can be reached by telephone and which are used in call centers
- intelligent systems that sort, answer, and forward incoming messages as well as emails used in customer support.
The limits of digital communication assistants
As a rule, digital assistance systems can do less than you might expect. It is therefore important to be aware of the limitations of these in order not to be disappointed when using them. Important boundaries include:
- Acceptance – Bots are often not accepted because many people do not feel valued by them. Past negative experiences also play a major role. Once you have had a bad user experience with a chatbot or voice assistant, you are less open to them in the future.
- Decision-making – Digital assistants do not know what is good or bad. They are therefore unable to classify different arguments and make decisions based on them. They can reproduce only the answers with which they were trained.
- Empathy – Bots are unable to empathize with the customer. They don’t notice when a person is angry or sad. They simply relay the answers unemotionally. Although they are excellent at providing numbers, data, facts, and answers, AI assistants are unable to recognize when someone does not understand the answer and then try to explain it in a different way. They would always give the same answer in this case.
- Context – Most assistance systems are currently unable to recognize the context of previous questions and answers. As soon as a user’s question is answered, the assistant starts from zero – as though the previous question had never existed. An assistant does not (yet) remember them.
Prerequisites for the implementation of digital communication assistants
In order to be able to develop a chatbot or voice assistant in the first place, several prerequisites must be considered. These are often not thought of in advance. The following therefore provides an overview of the most important points.
- What does the user really want? – First, you must find out why a user wants to make contact. Is it consultation in advance? Are they support requests – maybe even about one specific thing at a time? This classification is quite important and should be done first in order to get a feeling for the expectations. Here is a distribution of topics for which inquiries are frequently received and those for which inquiries are received only infrequently. You will see that there is a high demand for a complex of topics. You should then start with these topics when developing the digital assistant.
- Complex dialogues – When starting to develop a chatbot or voice assistant, you might not even be aware of the complex dialogues that can arise. There are countless branches, response options, and loops; these must be worked through and designed. Once the user’s expectations have been clearly defined, the journey through the entire dialogue continues. Although this sounds quite simple at first, it can turn out to be quite extensive depending on the particular request. It is important to put a lot of work into this phase because it is the basis for everything else. Better one iteration too many than one too few.
- Content is more important than technology – In the technology-driven environment of AI, this point is often neglected. However, it is one of the most important points when it comes to developing a good chatbot. Without good content and answers to specific questions, even the best AI can be stumped. It is therefore essential that a content library in which all possible dialogues are mapped and deposited with corresponding content be created in advance. The basis for this is the dialogue overview of the previous step. Companies that have prepared interview guidelines for their employees have an advantage. As a rule, these are a solid basis for the bots.
- Flexible data connections – A lot of content and complex dialogues require flexible data connections. By developing and maintaining of knowledge graphs, the complex data are connected with each other, and the bots can thus acquire new knowledge.
Current use cases for digital communication assistants
Below we describe a few scenarios in which intelligent assistants are used. Of course, this list does not claim to be exhaustive. After all, there are (almost) no limits to your fantasies.
- Fully automated call center and customer surveys
- Call center automation – By transforming questionnaires into human-like dialogues and automated calls, customer feedback responses can be collected in a structured manner.
- Reduction of calls by service workshops – By switching customer satisfaction surveys from calls to SMS, messenger or email, such repetitive tasks can be automated.
- 24/7 support and service for your customers
- Tips and support at Zoey – Zoey’s mix of chatbot and human interaction ensures that travelers get to their destination faster and can get the most out of their holidays because they are supported by the assistant on their smartphone.
- 24/7 shopping and type advice through an integrated chatbot on a fashion e-commerce store – Based on the preferences entered by the customer, the bot suggests a number of garments but also learns from the rejected products and thus iteratively improves its suggestions.
- Employees become more productive through digital assistants
- Digital assistants for employees – Chatbots integrated into a company’s communication channels (Slack, Teams or other chat systems) can relieve employees of unproductive tasks. From booking a meeting room to coordinating meeting participants to reserving a parking space.
- FAQs and internal communication – There are also chatbots that serve as staff information systems. For the distribution of internal messages and as assistance in answering FAQ, these bots can replace humans to a considerable extent.
Opportunities of digital communication assistants
In addition to all the limitations and prerequisites, digital assistants offer enormous potential for the future:
- Analysis and reporting – Bots are an excellent source of data on user interactions and needs and should thus be used and analyzed accordingly. On one hand, to constantly improve the algorithm. On the other hand, to get to know customers and their wishes better.
- Scaling of repetitive tasks – Routine tasks and questions can – and should – be taken over by digital assistants. However, as soon as things become more complex and personal, there needs to be a seamless transition to a human taking over the conversation. Humans can then respond to the specific needs of the users.
- Voicification – After the keyboard, the mouse and the finger, speech is the next interface in human-machine interaction. In the meantime, technology has come so far that it can already understand us humans quite well. The next step is to find suitable use cases that make interaction through this interface easier for humans.
- Context is key – In order for chatbots to have an increasingly human character, it is important that they recognize contexts and not simply recognize each question as being self-contained. If you ask where your appointment is tomorrow, you will get an address as an answer. If you then ask what the weather will be like there, almost all the assistants currently available will be unable to provide a valid answer because they have already forgotten the location. But this will change drastically in the future.
Have you already thought about possible areas of application in your company? Have you gotten initial ideas while reading? Or are you simply overloaded with information because of the numerous possibilities? No matter where you stand, we would be happy to support you in your decision-making and implementation regarding digital assistance systems.
Please feel free to contact us or inform yourself in advance, especially on the topic of return on investment for digital communication assistants. Because developing good chatbots requires a certain investment. In addition to the development of the dialogue patterns and the training of the AI, you may also have to put effort into creating the content. In order to give you a better indication, we have carried out exemplary return on investment calculations for several use cases. These will be published shortly. Register now so that you will be notified when they become available.