The BBC’s Beeb shows why AI voice assistants are not yet ready for enterprise prime time

Opinion Last week, the BBC announced that it has begun work on a voice-enabled virtual assistant scheduled for 2020. Codenamed Beeb, the new virtual assistant is being seen as an alternative to Amazon’s Alexa and Google Home, with a focus on helping users find content on the British broadcaster’s services and being able to understand British regional accents.

This last part is important, because it indicates that even as far as we have come with voice technology, companies are still stumbling at the first hurdle - accurately understanding their customers. Anyone who has ever interacted with a voice assistant will tell you that the likelihood of it misinterpreting a command is commonplace - doubly so, if you have a thick accent or are not a native speaker of its default language.

With chat, these basic understanding challenges are less prevalent. Advances in conversational artificial intelligence (AI) now make it easier than ever to understand slang, dialects and even typos. This allows users to communicate in a manner they feel comfortable with, rather than having to rely on remembering a series of predefined commands.

In the enterprise space, this level of understanding is especially key, because, as we all know, customers do not like to be kept waiting. Paying off your credit card balance via voice assistant only to have it misunderstand your request and add an extra zero is not a good customer experience. That’s not to mention the security implications that surround having your financial information easily accessible by voice. 

It’s far easier, and often more secure, to hop on to private, authenticated chat with a virtual agent and complete your transaction that way. All the info you need is displayed in the chat log, and there’s no need to repeat yourself multiple times to get what you want.

Best in class conversational AI solutions offer advanced language understanding functionality (some common and others proprietary) to make sense of the many permutations that words can take. This includes stemming, compound splitting and spelling correction, as well as semantic understanding. In a grammatically complex language, such as Finnish, for example, there are hundreds of conjugations for something as simple as ‘car insurance’ making it incredibly difficult to parse without access to the right technology. 

Conversational AI, through chat, can make easy work of even the most complex languages, whereas voice assistants have to contend with regional accents on top of these already difficult language challenges. 

That’s not to say that it will always remain this way. The potential of voice technology is certainly far-reaching and the Guardian reports that around 20% of British households already have some form of smart assistant. For now, however, voice in enterprise should be relegated to very specific cases that leave little margin for error or complexity. Building a robust foundation of conversational AI through chat should be a priority, so that companies can be exactly where their customers are; as a far higher percentage of consumers have access to chat via their smartphones, than they do to voice assistants.

When the understanding capabilities of the technology fully mature to a level that can compete with chat, the companies that have invested in laying the conversational AI groundwork will be able to more easily layer voice on top. Only then, can they deliver a truly next-level customer experience through voice and not one that has customers hurling small plastic discs across their living rooms in frustration.

(c)iStock.com/mikeinlondon

Read more from sister publication IoT News: The BBC plans to launch its own virtual assistant next year 

Interested in hearing leading global brands discuss subjects like this in person?

Find out more about Digital Marketing World Forum (#DMWF) Europe, London, North America, and Singapore.  

Related Stories

Leave a comment

Alternatively

This will only be used to quickly provide signup information and will not allow us to post to your account or appear on your timeline.