Increasingly human-like responses from ChatGPT and other large generative text models have employees and politicians concerned about impending waves of job losses. Meanwhile, contact centres are having trouble recruiting and retaining enough agents to keep up with increasing consumer demand. Is the stage set for a showdown between humankind and machine, is a stalemate more likely, or will we team up by hook or by crook?
Out of necessity, whether it be consumers voting with their feet or by regulator intervention, the way companies respond to (or ignore) their customer queries will change radically. Many companies have made a cost-cutting business decision to block customers from reaching a person in help centres, and others simply have inadequate customer services operations. Either way, the result is that businesses often frustrate customers in search of help. This bad customer experience comes with real costs: negative promotion, customer churn and missed revenue opportunities. It then puts agents in the uncomfortable position of facing unnecessarily vexed customers, compounding the difficulties businesses are facing in staff retention.
It's hard to be human
Stress and unreasonable demands mean many agents are in an untenable situation. They’re expected to juggle several chats at once and to close calls quickly. Most customers expect agents to know who they are and be aware of any previous issues or interactions. But agents often can’t easily access that type of information, which adds to the stress of the job. After months of training, agents still aren’t equipped to handle many of the cases they come across without help. Even with more experience they’re often powerless to solve the business process issues they face. Small wonder that many leave soon after starting.
Could new technology save the situation?
Large Language Models (LLM) like ChatGPT can feel like the magic solution and are capturing the imagination of businesses and consumers alike. Who wouldn’t want an all-knowing entity to solve all your problems?
Talk to Bard or ChatGPT at length though, and the flaws begin to show. They have a predilection for lying, lack contextual understanding, access outdated information, and for businesses there are security concerns. Despite the risks and pitfalls of Generative AI models they’ve reignited imaginations in the customer service space and can, with the correct application, support better interactions.
Bots alone can’t cut it
Even before the new generative AI was unleashed on the world, integrated, outcome-based bots had evolved to handle high-volume requests, but needed to handover to a human for anything that falls outside what they were built for. That’s still going to be the case, with or without generative AI. Generative AI understands requests better, which can make a huge difference in accomplishing a task. However, with responses that can’t be trusted, companies shouldn’t let them answer customers directly. Generative models can take a first pass at answers, but any responses they generate should be checked before use.
It’s not sexy, but generative models make it feasible for companies to gain full insight into what customers most frequently ask for, along with how they ask for it and the corresponding agent responses as well. Automating the bulk of categorisation and labelling work and learning the words and phrasing customers are using enables training of company NLP for better recognition. LLMs are great at figuring out alternate phrasings to further improve recognition rates.
Better together
Rather than placing demands on agents to constantly juggle more work faster, there are many ways that Conversational AI platforms can support, not replace them, in their work. Whether up front or behind the scenes, adding bots the right way can lighten agent workload.
Agent Assist technologies provide agents with suggestions to help customers, shortening agent training and response time, increasing their confidence and performance, and improving their job satisfaction.
Agents have the deepest insight into where business processes are going wrong and if they leave, that insight goes with them. Enabling them to improve their workplace can help to retain them and capture the value of their experience.
Becoming a bot trainer can be an attractive career progression path as well. Helping agents to improve bots catalyses a virtuous circle. It’s worth ensuring they understand there’s plenty of work for them for the foreseeable future – and that they can help shape their workplace into a place where they will thrive.
New tech isn’t the solution, but it is part of it
Contact centre interactions are symptoms of underlying business issues; paying attention to these signals can help you react and take steps to solve them. Thoughtful implementation of new technologies to support staff and reinvent business processes set you up for success, not just survival.
About this author
Cheryl Allebrand
Senior Consultant, specialising in Artificial intelligence (AI) and Automation
With close to two decades of experience in tech and strategy, Cheryl is dedicated to finding solutions that work for organisations, their members and those they support.
