Voice

Let’s not get carried away as we keep using virtual assistants

Different studies show that one in four Germans already owns a smart speaker. Since the infrastructure is basically there and language technology is constantly evolving, we can indeed speak of "The Next Big Thing", argues Nicolas Konnerth – thus reacting to a sceptical comment by Markus Sekulla.


According to Gartner’s Hype Cycle, every trend follows a similar pattern: an innovation trigger is followed by a peak of inflated expectations, and then the trend falls into the trough of disillusionment – before ultimately and notably achieving long-term, successful growth on the plateau of productivity. As the peaks and troughs of this cycle are more or less pronounced depending on the trend concerned, it is always difficult to identify where a trend seems to be at any given time. But when it comes to virtual assistants, I would go out on a limb and say that we have moved far beyond the peak of inflated expectations. Various studies indicate that one in four people in Germany owns a smart speaker, which is a voice-controlled virtual assistant with a loudspeaker. According to ABI Research, some 20 million new motor vehicles will be pre-configured for popular virtual assistants; drivers will not have to do without their modern digital friends. In other words, the infrastructure already exists.

But the question remains: how many people regularly use virtual assistants? Google’s Sindar Pichai announced during a keynote as long ago as 2016 that about 20% of queries on mobile devices were voice searches. One in three Germans now regularly uses a virtual assistant.   

How are virtual assistants being used?

In short, virtual assistants are clearly being used by the general public. But for what purposes? I must admit that Markus Sekulla is right when he points out that virtual assistants are most frequently used for pretty trivial things, including communicating, navigating, listening to music, retrieving basic information such as the news or a weather forecast, and controlling smart-home devices.

People expecting more meaningful dialogues or something closer to humanity are perhaps stuck atop the peak of inflated expectations. This is not the fault of the virtual assistants; after all, we cannot expect any “assistant” to be capable of everything. They are nevertheless always eager to aid us, simply awaiting spoken commands from users – most of whom do not mind that the assistants are not perfect. This explains why a lot of people own one. 

The hope for increasingly advanced conversations

We can remain hopeful, of course, that virtual assistants will increasingly advance beyond constrained exchanges and engage in more intelligent conversations. Just such a future is not far away; in fact, it has already begun. Google demonstrated this very impressively back in 2018 with its Duplex, an AI system for conducting conversations. As requested by a user, this system called a hair salon and scheduled an appointment. The use case itself was not remarkable, but the manner in which the virtual assistant made the appointment certainly was. Duplex successfully imitated a human being, used filler words such as um and uh, and even asked for a moment to check the appointment in its calendar. The illusion was so convincing that people started asking whether Google Duplex had rather effortlessly passed the legendary Turing test.

Alan Turing, the famous computer scientist, predicted in 1950 that by 2000 artificial intelligences would be able to fool humans so well that people would have no more than a 70% chance of being able to distinguish between AI and human interlocutors. Strictly speaking, the conversation carried out by Google Duplex did not pass the Turing test. The topic of the phone call was too constrained and the woman who spoke with Duplex was not aware of her role as guinea pig. All the same, the exercise does nourish hope for increasingly intelligent virtual interlocutors. 

Next Big Thing...

Google Duplex ceased being a gimmicky novelty long ago – people in the United States can already use it in combination with a second service that handles online bookings, for example.

“But what does that have to do with my Amazon Echo or my Google Home?” is a question some readers might be asking themselves. To which I would respond: “A great deal!” The refinement of voice-activated technologies lies less in hardware and more in the cloud. Every owner of a smart speaker will benefit automatically from the continual advances in voice-activated technology – without needing to bother with annoying or expensive hardware upgrades. Amazon now has more than 10,000 employees dedicated to making Alexa even more intelligent. Considering that the infrastructure basically already exists and that voice-activated technology is improving all the time, we are indeed talking about the next big thing.

Text: Nicolas Konnerth

What does the Head of Voice at ERGO do? (Video in German only)