The rise of omni-channel customer experience (CX) means there have never been more ways for consumers to engage with the brands they love, or hate. Many new digital and social channels have burst onto the scene in recent years, though some have survived longer than others. Chatbots have notoriously struggled to impress, with only 3% of customers considering them their first choice of support channel. The era of chatbots is far from over, however. Can chatbots recover their mojo by riding the wave of positive sentiment created by Large Language Model (LLM) generative Artificial Intelligence (AI) technologies such as GPT-4?
Chat-based technology is not a new concept, and it is likely you will have already used both live chat and AI-generated chatbot technology, known as conversational AI, when reaching out to a business for support. Chat capabilities have been welcomed by businesses, but consumer attitudes are less straightforward. Over one-third of consumers consider live chat options, operated by human agents, as their first choice when reaching out to a brand. This human-chat-preference proportion far outweighs the numbers of those choosing to use chatbots in the first instance and highlights the current disparity between AI and human-led conversation. Many consumers have had at least one poor experience with a chatbot. So what are the limitations holding chatbots back?
Customers look to resolve issues through the most efficient channel
While traditional chatbots are automated, they are incapable of adapting to more complex queries, or answering less traditional questions. This means that customers with nuanced enquiries are, more often than not, left dissatisfied by their chatbot experience. Chatbots also traditionally struggle to provide the empathy desired when it comes to urgent enquiries, for which human customers almost always prefer to deal with human agents. Conversational AI goes a long way towards making interactions more effective; but first customers must give the bots a second chance, which is no small ask. The hype surrounding OpenAI’s ChatGPT, and its successor, GPT-4, presents an opportunity to rewrite the narrative around conversational AI.
The Chat-GPT buzz
ChatGPT burst into the public eye and caught the imagination of the masses. The free-to-use version of ChatGPT allowed the public to test the potential of LLM-based generative AI and to be inspired by the breadth of its ability. Buoyed by intensive media attention, fuelled in part by the high-profile backers of the technology’s owner, OpenAI, ChatGPT became an overnight sensation and almost a household name, with commentators prophesying an AI revolution. The rapid release of GPT-4 added even more fuel to this fire. Almost instantaneously, the public perception of AI changed. No longer clunky and confusing, now modern, flexible, and with limitless potential to solve problems, pass exams, and maybe even create mass unemployment. If customer service chatbots could offer interactions incorporating such advanced technology, then customer experiences would become effortless, and user hesitancy would evaporate.
The modern chatbot
So what does modern conversational AI do differently? Rather than deviating from the core principles of a chatbot, solutions such as ChatGPT build on them. Early chatbots relied solely on Natural Language Processing (NLP) technology. NLP breaks text down into its component parts to discern meaning. Whilst this works for simple enquiries, it can struggle with more complex sentences and intentions, leading to confusion and frustration for the customer. In contrast, today’s AI chatbots are built around Natural Language Understanding (NLU). NLU employs deep learning to identify sentiment, topics, and intent. By drawing on a broader knowledge base, NLU allows the chatbot to support its answers with understanding of the underlying data and context. Additionally, NLU allows the chatbot to respond in a human-like way, tailoring the response to the user, especially when integrated with a Customer Data Platform, which is the newest iteration of the Customer Relationship Management (CRM) concept, but spanning multiple systems of record.
Unsurprisingly, ChatGPT ‘believes’ that it can improve CX. When asked about its own potential, the AI responded with some thoughtful insights:
“Conversational AI models like ChatGPT are likely to improve chatbot technology for customer service in the future. These AI models have the ability to understand natural language, generate human-like responses, and learn from conversations they have had with customers. However, it’s worth noting that while conversational AI models like ChatGPT have the potential to improve chatbot technology, there is still a lot of work that needs to be done to make these systems more effective and efficient. For example, chatbots need to be trained on a large and diverse dataset of customer interactions to ensure they can handle a wide range of customer queries and requests. Additionally, chatbots must be able to identify and handle complex or sensitive issues that may require human intervention.”
Despite ChatGPT’s enthusiasm and self-confidence, there is still work to be done in order to make real-world progress. The contact centre occupies the front lines of the AI revolution, but overcoming user hesitancy will be vital to making the next big push forward. Whilst technology has advanced to the point where AI can play a vital role in CX, its human bosses will need to deploy the technology carefully and thoughtfully if they are to meet, and exceed, ever-greater customer demands with their newest chatbots.
