.png)
With customer service in crisis and chatbots on the rise, find out how gen AI is becoming the new face of customer care for growth-orientated organisations.
Key takeaways:
- Generative AIs like ChatGPT that employ Natural Language Processing (NLP) Models are on the rise.
- They hold revolutionary potential for transforming several business operations, including customer service.
- Benefits include drastically reducing time-on-tasks, heightening personalisation and deploying advanced vulnerability analysis.
- Some organisations still resist adopting gen AI, with critics focusing on emotional intelligence being unique to humans.
- But with forward-thinking companies paving the way, this new tech is here to stay.
The public release of OpenAI’s ChatGPT in November 2022 was a watershed moment for generative AI.
A powerful Natural Language Processing (NLP) model, with the ability to churn out everything from essays to OpEds in seconds, ChatGPT became an overnight household name.
Over the years there have been many cycles of hype and scepticism for AI, but ChatGPT felt like a turning point. For the first time, conversations about the competitive potential of NLPs and gen AI were being seriously discussed in virtually every boardroom globally.
A Silicon Valley-style arms race even ensued, with tech firms dashing to develop their own adaptive models ahead of competitors.
Despite the initial furore, two years on (and two ChatGPT updates later) gen AI has only been tentatively adopted by most organisations for operational automation, data processing and the occasional, often badly written, blog post.
So why, in the face of pioneering applications — such as the ability to radically transform entire customer service and communication operations — are companies still throwing caution to the wind?
Resistance, for most, seemingly hinges on one thorny point — emotional intelligence. Can gen AI models really emulate an emotionally sensitive, empathetic customer experience?
When you add in the potential for factual inaccuracies and regulatory risk, do human agents, as the critics would contend, still reign supreme?
The state of the struggling customer service industry
Forbes recently reported that bad customer service experience could be costing the global economy as much as $3.7 trillion a year. In other words, bad customer service is bad, bad business.
Since the pandemic, rising financial demands have led to dramatic cost-cutting measures for many firms, with customer service operations being one of the hardest-hit departments.
Many businesses have hugely scaled-down staff, both internally and with outsourced providers, reducing their overall capacity for dealing with customer inquiries. Despite slimmed-down teams, since the cost of living crisis, businesses are seeing record levels of complaints and inquiry demands.
The result? Arduous customer wait times, poorly dealt with complaints, burnt-out staff and high customer churn.
To put this into perspective, the average wait time to get through to a customer service agent in the UK during the quietest period is 10 minutes and 7 secondsCall up an energy provider and you’re looking at more than a 35-minute wait.
Gen AI adoption or not, the state of customer service is in crisis.
Gen AI: the automation taskmaster
Wait time bottlenecks often stem from customer service agents having to carry out lengthy manual tasks, even within well-scaled teams. Gen AI, however, can revolutionise almost all of these time-intensive processes.
Take for example a previously labour-intensive task like reading through and deciphering a lengthy interaction history with a disgruntled customer, and then writing a suitable message to resolve the conflict.
With a simple prompt, gen AI can sift through vast amounts of information and data from multiple sources, summarising an entire interaction history in a fraction of a second.
And it doesn’t stop there. It can then simultaneously use that data to create a personalised message that responds and adapts to that particular customer's conversational style and behavioural cues.
Throw in additional tools like automated note-taking and next-action troubleshooting, and human agents are freed up to spend more quality time with customers, rather than spending their days multitasking, writing repetitive messages or sifting through customer data.
The output isn’t just drastically reduced time-on-task, but heightened personalisation and in turn an enhanced customer experience.
Klarna, the AI-powered global payments network, stated earlier this year that up to 66% of its customer service chats are now handled by an AI assistant.
Powered by OpenAI, Klarna's AI assistant handled 2.3 million conversations in its first month. That’s the equivalent workload of up to 700 human customer service agents. Klarna estimates that this will translate into a $40 million profit uplift in 2024.
Can generative AI emulate empathy?
So AI can generate a personalised message to resolve a conflict, but what about emotional intelligence? Can your friendly neighbourhood chatbot dapple communications with the necessary empathy, understanding and cultural sensitivity they need to land just right?
Naysayers contend that no, gen AI can’t emulate the human-level emotional ingenuity crucial for dealing with complex customer interactions. This is predicated on the fact that gen AI can’t ‘think’ for itself, or employ emotional reasoning.
This, in part, is true. Gen AI can decipher, sort and then recite learnt information exceptionally well, but it does so within the limits of predefined parameters. This means anything it generates will essentially always be a variation of the content it was trained on.
But what happens when that content includes specific tone of voice cues, behavioural science principles, nuanced cultural references, past interaction histories and hypersensitivity to customer vulnerabilities?
The outputs, in fact, become staggeringly effective. Empathy, let’s not forget, is a learnt behaviour.
Another concern from critics has understandably been about the quality of the text gen AI can produce. Within the context of customer service, this concern also extends to whether the content can ‘pass’ as human-generated.
But with rigorous and factually correct training data (unlike ChatGPT which draws heavily from Wikipedia) in addition to some of the prompts and parameters mentioned above, the quality of the content can be tightly controlled.
When responding to inquiries over email, outputs can look and feel strikingly similar to human-generated content, especially when models have been trained on large datasets of human-written replies.
One company paving the way for AI-generated customer communications is Ophelos’ client Octopus Energy. Now the UK's second-largest energy provider, Octopus answers up to 45% of customer queries with generative AI alone.
Founder Greg Jackson recently commented “We think computers are not as empathetic as a human, yet the very first email I saw that had been written by gen AI was to a customer who was struggling to pay their bills because of ill health. [It] opened by saying: ‘At times like this, there are more important things to worry about than your energy bills. First and foremost, I hope your health has improved’.
Mitigating risk in real time
Whilst generative AI alone can enhance customer care, it can be elevated even further when paired with additional models, such as NLPs.
When trained using highly sophisticated datasets, NLPs can be used to automatically detect emotional cues and conversational patterns across multiple issues. Gen AI is then able to collate these trends and produce a reactive response or flag certain risk parameters to human agents.
At Ophelos, we developed OLIVE, a Natural Language Processing (NLP) model that uses deep learning and conversation cue detection to predict each customer’s vulnerability risk.
OLIVE can classify 24 different vulnerabilities – it scans every customer interaction, rating the customer risk profile for different issues as low, medium or high. This enables customers with higher risk profiles to be flagged to human customer service agents, with smaller, lower-risk inquiries being handled directly by our generative AI model, OphelosGPT.
The average response time for OphelosGPT is 3 seconds – meaning lower-level inquiries such as login issues, payment updates and account management queries are resolved almost immediately.
Any potential safeguarding issues are reported to agents and prioritised in real-time, rather than left piling up in an inbox until an agent is available. This also has enormous benefits for compliance and regulatory risk mitigation – which is business-critical within the context of debt collection and the wider financial product landscape.
Both OphelosGPT and OLIVE have in-built compliance parameters, allowing for any potential issues to be flagged in real-time and all AI-generated content to be 100% compliant with regulation updates – all without needing to train any staff in new compliance procedures.
AI-generated content also drastically reduces the likelihood of human error, which is now the most likely cause of data breaches and cyber incidents globally.
The future of customer service: AI way or the highway
Exemplary customer interactions are built not only on empathetic, human-feeling communications but also on personalised, effective dialogues that ultimately provide the customer with a timely resolution to their query or situation.
Writing off AI as unable to emulate learnt behaviours like empathy, understanding or cultural sensitivity feels reductionist.
Gen AI is transforming how high-growth organisations are approaching their customer service, both from the perspective of their bottom line and the overall experience they’re providing their valued customer base.
When applied at scale, with potentially an infinite number of complex responses, it’s easy to see gen AI’s superior use case over any regular predefined digital journeys or Interactive Voice Response (IVR) systems.
With the rise of Gen-Z consumers, where one bad tweet can tank a company's share price, positive customer care interactions are becoming increasingly integral to how businesses are perceived by the market.
Forward-thinking organisations are adopting this evolving technology, and those who don’t are quickly in danger of being left behind.
Intrigued to see what gen AI could do for your customer service operations? Get in touch today.
