That’s what language dialogue systems can do today – and that will bring the future

LEAD: Should one be polite to a language system?

Georg Theunissen: This is not necessary. Above all, it is important to clearly state what it is about and what you want. So “I want to book a flight to Majorca” is good, but “I want to book a flight from Frankfurt to Mallorca next Saturday morning and the return flight on Sunday night” is better because the system can make specific proposals without further queries.

Whether one adds “please” or not, is indifferent to the matter. Excessive courtesies such as “how are you?” Should be avoided, as they only distract and confuse the computer.

LEAD: Why does it even have advantages to speak with a language system and not with a flesh-and-blood person?

Theunissen: Many callers find the anonymity of a voice computer more enjoyable than talking to a human. In addition, the language system is always equally competent because it can access all relevant data immediately.

On the other hand, there is a wider range of experiences in a man-made call center. You may get to someone who can not help because of lack of experience, even if the more experienced colleague sits right next to it. With a language system, you always have access to the best available competence.


LEAD: Should I share sensitive information such as finances or medical findings with a language system?

Theunissen: Yes, without further ado – after a security inquiry has been made. This can be a PIN request or a password, or of course – most certainly – voice authentication. Then you can entrust the computer with everything that is required for the task. Safety is many times higher than when a person is on the other end of the line.

You might also be interested in this: Siri and Alexa – Why are language assistants female?

LEAD: How does that work with voice authentication?

Theunissen: To do this, one has to identify oneself by means of one’s voice and in all the coming cases will be recognized by the voice alone. So the first time you enter your PIN code – or whatever you’ve previously identified – and then participate in voice authentication, which takes less than a minute.

In the future, the system will recognize you as soon as you start talking. Important: Modern voice authentication systems continuously check the voice throughout the call. As soon as another speaker appears, the process is aborted for security reasons.

LEAD: Will there ever be only some language systems in certain areas?

Theunissen: In any case. As early as 2020, the first voice dialog systems will come into operation, in which the caller will no longer recognize at least in some cases that he is talking to a computer. In 2025 this should already be the case for most calls. In all these cases, the voice computer is faster, easier and, of course, cheaper. So there is no reason to use a human for this.

You might also be interested in this: Study on the use of language assistants

Georg Theunissen is sales and consulting director of Swiss Spitch AG. (Photo: Theunissen)

LEAD: How have the language systems evolved in recent years?

Theunissen: The greatest progress has been made in two areas: voice recognition and dialogue. These two factors will also be decisive for future development over the next few years. There is also another factor: artificial intelligence.

The more systems are equipped with AI, the easier it is to identify requests and respond to them in a targeted way. The combination of better and better voice recognition, voice output and dialogue guidance thanks to Artificial Intelligence will mean that in the future you will only be aware of whether you are talking to a computer or to a human being in exceptional cases.

LEAD: How much can a company save by using voice systems?

Theunissen: Today’s systems come up with cost savings of between 30 and 50 percent, in some cases as much as 70 percent. This percentage will continue to increase in the future because technological development will lead to such perfection that the systems can handle an increasing part of the requests themselves instead of putting them through to a human being.

LEAD: In your experience: Do customers prefer talking to machines or people in 2019?

Theunissen: Since many still have rather bad experiences with legacy computer systems, such as “say 1 for a new contract, 2 for a service request”, most prefer one person as a contact person. However, this only applies until you have communicated with a modern speech dialogue system.

Once you get used to the new kind of language system, you do not want to miss it any more: the computer colleague picks up immediately, stays always friendly and above all always competent and can therefore help much faster in almost all cases than the human being.

Leave a Reply

Your email address will not be published. Required fields are marked *