While artificial Intelligence (AI) technologies have been steadily evolving since the 1960s, latterly including image recognition and natural language processing (NLP), 2023 will be remembered as the year that generative AI (gen AI) burst onto the scene. In fact ‘Artificial Intelligence’ and ‘hallucinate’ have both entered common parlance after being named ‘words of the year’ for 2023 by Collins and Cambridge Dictionaries.
Released in November 2022, OpenAI’s ChatGPT immediately captured the public imagination, quickly becoming a popular byword for gen AI and even AI in general. ChatGPT’s emergence heralded a cascade of similar Gen AI tools from competitors. Thanks to its popularity and accessibility, 45% of Contact Centre-as-a-Service (CCaaS) executives say that ChatGPT prompted a rise in AI investments within the Customer Experience (CX) industry, seen by many as “ground zero” for the adoption of Gen AI by mainstream business.
If 2023 was the year that gen AI captured the world’s imagination, 2024 will be the year that we see its first mainstream deployments, starting with CX. What were the milestones of gen AI in CX during 2023, and what opportunities and obstacles lie ahead the coming year?
1. Intelligent customer journeys
Intelligent automation can make contact wait time productive, by integrating AI with CCaaS systems to streamline the customer journey. During an interactive voice response (IVR) process at the start of a call, or the beginning of a chatbot dialogue, AI can segment. How does this work? It collects digital contacts by automatically gathering data at the start of an interaction to identify acustomer and their intent.
Previously, service agents were forced to perform the repetitive processes of identifying and verifying customers, their mundane questions adding to the frustration of callers who had already been kept waiting. AI streamlines the “who are you and what do you want” process for agent and customer alike, enabling both parties to tackle the substance of the matter straight away.
2. Interaction summarisation & notes
Another example of AI-enabled technologies being embedded into the contact centre is the automation of call notes and summaries. A sizeable fraction of an agent’s time is spent wrapping up interactions, and now gen AI can take care of that legwork.
NLP can transcribe the interaction, which can then be translated and summarised by gen AI as appropriate, and customised into different types of forms such as a purchase order or a complaints process, which the agent can then check and amend if needed. With AI doing the heavy lifting, agents will have less admin to do, so are free to engage in dialogue, solve problems, and build trust with the customer, rather than the traditional “talk and type” approach.
3. Generating knowledge articles
10-15% of an agent’s time can be spent looking up information to assist a customer, who is often left listening to poor-quality audio of Mozart and feeling devalued. Knowing where and how to search all the obscure tabs in a contact centre’s back-end systems forms a vital part of agent training time, which still takes an average of 12 weeks (and in an industry where 50% annual headcount churn is commonplace).
Gen AI, whilst in its infancy, is set to make even more inroads into the CX space in 2024, with more sophisticated applications made feasible. These will include automatically-created knowledge articles based on the context of an interaction, placed on the agent’s screen as a conversation evolves, and providing the agent with the information they need, along with relevant sources tohelp answer a customer’s query in real time.
Navigating hallucination in the CX landscape
This year will see gen AI face its biggest challenge yet; the battle against hallucination. OpenAI, the leading AI research and deployment provider, is transparent about the fact that around 1 in 5 of ChatGPT’s responses include incorrect information. A 20% hallucination rate could see hundreds if not thousands of customers misinformed every day when communicating with a brand. The potential consequences of sharing inaccurate information should not be underestimated. Brands may quickly lose customer loyalty if they are given incorrect information during interactions, and a tool meant to improve First Contact Resolution(FCR) could instead have the opposite effect.
The threat of gen AI hallucination can be drastically reduced by narrowing down the volume of information against which the AI is trained, to only include the organisation’s own information and a limited number of other trusted sources. Agents will also play a vital role in the fight against misinformation caused by AI hallucinations. Agents must have access to properlyreferenced sources, so they can manually verify check the source validity of AI outputs.
2024 and beyond: what comes next for gen AI?
The Gartner Hype Cycle for 2023 predicted that mainstream adoption of ‘emergent’ AI technologies, such as gen AI, will begin as soon as 2025. As we move into 2024, we will likely begin to see use cases of AI spread across CX, before, during and after interactions including automated customer satisfaction surveying and AI-conducted audits. AI will be used to train employees to perform like skilled veterans in record time, using data-driven insights.
The CX dream of hyper-personalised interactions on a population-wide scale, driving unheralded levels of workforce efficiency and productivity, is now closer than ever to being realised.