Artificial Intelligence (AI) The Next Big Thing?
A great deal has changed within Customer Care since the introduction of the first Automatic Call Distribution (ACD) in 1970s by Rockwell. We have moved from voice centric Call Centres to Contact Centres, with the addition of more processing and channels. Omni-channel further extended the channels available to customers and Agents to Connected Digital Experience.
The ‘Connected Digital Experience’ truly reflects Customer Care when we are ‘Connected’ by machine to machine, people to people or machine to people; ‘Digital’ with the wide range of channels and channel options and ‘Experiences’ as this now drives customers outcomes.
Machines now have the ability to have natural conversations as developers build on machine intelligence and learning. The advancement in cognitive learning now allows machines to recognise pictures, objects, speech, text, gestures, video, faces and even emotions.
The ability to have non-linear conversations is a real game changer, with the next generation of AI being both factually and emotionally aware, thus being able to provide a personalised service.
In the context of Customer Care, the most important function of any AI is its ability to understand natural language. The ability to do this requires the understanding of Natural Language Processing, understanding and generation.
Natural Language Processing (NPL) is focused on the natural human language integrations between computers and people. It is the science behind computational linguistics and provides the AI capability to process and respond to language.
The history of NLP started in the 1950s, although evidence can be found from earlier periods. In 1950, Alan Turing published an article titled "Computing Machinery and Intelligence" which proposed what is now called the Turing test as a criterion of intelligence.
Natural Language Understanding (NLU) is the comprehension of a machine/computer to process natural language, to pull apart and understand the language.
Natural Language Generation (NLG) or language production acts as a translator to generate natural language from a machine. This is normally from a knowledge base or logical data set.
The challenge is to provide conversational AI; to understand complex conversions using the methods detailed above.
A whole industry has been born out of both conversational AI and conversational interfaces. There are a number of open source providers, as well as more established companies, offering conversational AI technologies. You would think that this is simple?
Natural Language Understanding remains elusive and it’s much more than extracting intent from knowledge bases. To build a good starting point you need to analyse large data sets, such as call recording, to have a greater understanding of customer conversations. Once you have this information, you need to carefully manage the context and the dialog to then generate a conversation. This allows the AI to answer questions, carry out commands and engage with the customer with a level of intelligence.
The challenge is pulling together all these moving parts to provide a more natural dialogue between machine and people. Many of the solutions provided today are binary and have a very basic ability to form conversations and provide requested information. I’ve spoken to some experts who have created and provided some, but not all, of the functionality to achieve conversational AI. I have no doubt that within the next 18 months we will start to see a more sophisticated use of AI within Customer Care, but is the customer ready to talk to machines?