With today’s ever-advancing modes of communications between consumers and their service providers, artificial intelligence (AI) is about convenience. It’s about delivering information in a way that it simple and effective in order to allow them to get on with their work and personal lives.
For example, within a bank, it’s about speeding up access to information and providing automated resolutions. In a hospital, it’s about providing doctors and nurses with additional, automated support to better treat patients.
In each of these scenarios, one of the forms of AI making this possible is the chatbot. Chatbots provide a digital means of communicating on websites and mobile devices (such as chatting online or through a mobile app, or talking to a computer over the phone), to make things easier, save time, and allow users to focus on the bigger picture. This is why so many companies have deployed or are experimenting with chatbots – the happier the customer, the stronger their bottom line.
In fact, Grand View Research forecasts the chatbot market will be valued at US$1.25 billion by 2025 – that’s a compounding annual growth rate of 24.3 per cent. The catalysts for this include the promise of reduced operating costs, as well as ongoing advancements in machine learning.
But, for the most part, today’s chatbots are not a type of AI at all – they are linear, scripted voice-to-text (or text-to-text) tools that perform a limited set of predetermined tasks. While they drive significant value in specific ways, they often hit brick walls because they aren’t programmed to manage new tasks. If we are to truly transform today’s chatbots into AI, we need to plug the gap. So what’s missing? Machine learning.
Most chatbots today are built only with Natural Language Processing (NLP) capabilities – they are restricted to a predetermined (albeit an extremely expansive) set of words, sentence, languages and accents, and create dialogue based around this.
If chatbots are to become commonplace, they need the ability to proactively source new data, analyse it, learn it and deliver responses based on this new knowledge. That’s where the real value lies – and that’s where a mere ‘chatbot’ becomes ‘conversational AI’. As with any big data, storing and relaying it can only be so valuable – analysing it is when it becomes a key resource.
In laymen’s terms, conversational AI is essentially a single domain chatbot. It’s made for a specific purpose – for example, a bank billing enquiries system. However, its differentiator lies in its ability to autonomously create an open domain conversation with a customer by leveraging machine learning to expand its knowledge base. This can range from tapping into different databases and bespoke applications, to proactively interacting with other chatbots to identify the correct data, and even leveraging blockchains. All of this happens within a secure, private network to ensure that information is protected at all times.
A large Australian financial institution is currently in the process of implementing conversational AI that will boost account security by combining machine learning and voice biometrics. As soon as a customer calls into a contact centre, the system will confirm their identity through natural conversation, with conversational AI analysing customer information to equip the company’s agents with all the information they need to resolve the enquiry as quickly as possible. This also paves the way to leverage blockchain in order to ensure that the financial data being discussed is authentic, particularly if there are instances of fraud.
Applying this type of system requires a granular understanding of an organisation’s objectives. When we speak to our customers, we develop a customer journey map which outlines customer touch points to establish the best strategies for enhancing customer experience.
The technology component is designed after those business needs are determined, at which point we take a model-driven software development approach to build onto our open platforms and APIs. We look at every stage of the customer journey to develop the appropriate strategy for each specific scenario. We can therefore enable our customers to better understand the consumers they engage with by incorporating intelligent conversational AI into their key processes.
While most chatbots don’t quite meet the AI mantra at present, organisations have the opportunity to generate real value for their clients and their businesses by exploring the role of machine learning as part of their technology roadmap. Once that hurdle is overcome, chatbots will transform into conversational AI which can deliver on the promise of convenience so that less resources need to be invested in minute tasks, leading to enhanced experiences for customers and employees.
Peter Chidiac is managing director A/NZ at Avaya.