Generative AI’s 6 Paradoxes for Customer Service
With generative artificial intelligence (genAI) chatbots in the contact center, the transformation opportunities are tantalizing. But realizing genAI’s full potential requires a nuanced view of the different dynamics at play. Most of the advice you see out there falls into the category of “this gee-whiz technology can work wonders.” To be fair, pitfalls of genAI like hallucinations are understood and mitigation steps are being worked out. But the tensions and undercurrents in the wake of genAI are not as well understood or discussed. A recent must-read paper, “The Paradoxes of GenAI-Enabled Customer Service: A Guide for Managers,” explores this topic.
The authors identify six paradoxes of genAI in customer service: the potential tensions between connection and isolation; cost savings and societal impact; high quality and low empathy; satisfaction and frustration; personalization and privacy; and power and vulnerability. Customer service leaders who understand and deftly navigate these paradoxes will be off to a great start on their AI transformation journey.
Let’s unpack the six paradoxes:
Connected yet isolated. By offering personalized service around the clock, genAI chatbots can make customers feel more connected to a brand. But the lack of human interaction can also lead to a sense of isolation, particularly for customers who value a personal touch. The challenge here is to strike a balance between AI-driven efficiency and human connection.
Consider a “bot plus human” hybrid model: GenAI chatbots can handle simple tasks and routine inquiries while transitioning to human agents for complex or sensitive issues. The trick here is to preserve continuity and context to make this hand-off seamless.
Lower cost yet higher price. The long-term savings potential of genAI is appealing because it can be scaled easily, deployed 24/7 without overtime pay, and handle requests and process information faster than human agents. But automation’s widespread adoption not only raises the specter of job displacement in the service sector but can lead to the erosion of intangible knowledge and human skills required for customer service success.
In this case, genAI implementations can prove to be pyrrhic victories: Who’s going to handle the complex queries in the end? You need to invest in reskilling and upskilling for human agents so that they can provide oversight of AI bots and also handle more complex and high-value customer interactions.
Higher quality yet less empathy. GenAI chatbots may provide high-quality responses consistently, which can result in faster resolution times and better-quality KPIs. But they come up short in commiserating with customer when there’s a need for empathy and understanding. Lacking such emotional intelligence, they are ill-suited for handling upset or frustrated customers or dealing with sensitive issues.
This paradox is best addressed by gauging customer sentiment and generating “empathetic” responses, but more importantly by identifying situations that require a human agent in the loop and roping them in accordingly.
Satisfied yet frustrated. The quick and efficient service of chatbots often satisfies customer needs and results in high customer satisfaction ratings. However, when they misfire by misunderstanding customer intent or provide non-sequitur information, the tables are turned and it can lead to customer frustration. It’s sunny one minute and pouring the next!
What’s the solution? Customers should be able to express their dissatisfaction by easy feedback mechanisms. AI should be trained to recognize signs of customer frustration, such as changes in tone and language; transfers to human agents can be accordingly triggered.
Personalized yet intrusive. GenAI brings us to the threshold of personalization at scale of sorts: Based on past interactions and preferences, the turn-by-turn dialogues can be tailored just for you! Customer experience can be taken to the next level, but always lurking are the niggling questions of privacy. Customers may be spooked or unsettled by the deep knowledge that AI (seemingly) has about them and find it intrusive.
The answer: Go for transparency. Be up front about your data collection and usage practices. Give customers control over their data; they should be able to opt out. Implement multiple levels of personalization options and let customers choose the level they’re comfortable with.
Powerful yet vulnerable. GenAI is highly capable but also vulnerable. It can generate inaccurate information and be exploited by malicious actors and plagued by the biases of training data.
This is a tricky issue to navigate. Safety and security measures, guardrails, and monitoring mechanisms are a must. Regularly audit AI-generated responses and recalibrate as required.
Delivering successful customer service requires a blend of efficiency but also human connection. By understanding these paradoxes, you can unlock both AI scalability and human empathy. x
Kashyap Kompella, CFA, is an industry analyst, author, educator, and adviser. He is the founder of the AI advisory outfits RPA2AI Research and AI Profs and is a For Humanity Certified AI Auditor.
Related Articles
AI-based voice apps can help in the early diagnosis of disorders such as Parkinson's and Alzheimer's.
09 Jul 2024
It is the best of times, it is the worst of times for this powerful technology.
14 May 2024
Humans rely on audio and visual cues to comprehend speech, and ASR should do the same.
13 Mar 2024