Bot or not?: The effects of humanizing AI in the contact center
By Nikola Mrkšić, co-founder and CEO of PolyAI
Call center voice AI has come a long way from prototypical interactive voice response systems (IVRs) with convoluted menu trees and the constant robot-voiced refrain of, “I’m sorry, I didn’t quite catch that. In a few words…”
In fact, some voice AI is so sophisticated that callers don’t realize they aren’t talking to a person. This powerful technology could allow contact centers to automate many customer interactions with AI while still providing a highly engaging experience. But here's where it gets muddy — and potentially controversial.
As synthesized voices become more human-like, a critical question arises: Should customers be informed upfront that they're speaking with AI? On one side, transparency advocates argue it's essential for maintaining customer trust. On the other, some believe that overtly robotic experiences might remind callers of frustrating past encounters with less advanced technology.
Companies must be diligent about developing AI focused on helpfulness and unbiased service — not deception. Robust AI governance and human oversight are critical. With the proper safeguards, conversational AI can benefit companies and customers alike through more efficient, high-quality interactions that get issues resolved so effectively, the fact the agent isn’t human fades into the background.
The evolution of voice AI in the contact center
A history of ineffective IVRs has forced many callers to expect a less-than-ideal experience when reaching out to the contact center. For some, hearing a robotic voice on the other end of the line triggers an immediate exclamation of, “Speak to a representative!” in an attempt to bypass the bot entirely because they don’t believe it will provide them with the level of service they seek. In fact, according to recent research, 81% of people would rather wait for a live agent than immediately speak to a bot.
For others, hearing a robotic voice changes the way they speak — some callers will slow their speech and simplify the language they use in an attempt to “help” the bot understand them. However, these changes in speech patterns can sometimes make it more difficult for the bot to understand the caller, resulting in the dreaded, “I’m sorry, I didn’t quite catch that…” response. Pretty soon, these callers will ask to speak to a representative, and so continues the cycle. Because of the precedent set by limited automated capabilities, callers think the worst at the get-go.
Recently, however, improvements in conversational AI tech, large language models (LLMs) and subsequent generative AI options have changed the game for call centers. Leading call centers have moved away from ineffective, mechanical voices to speech that sounds so natural many users may not even realize they’re speaking to a voice assistant. Additionally, because of advancements in speech recognition and comprehension abilities, callers can speak to the voice assistant naturally, meaning the caller can use slang, pause or ask multiple questions at a time and the voice assistant will still be able to understand. The more customers speak to these bleeding-edge AI voice assistants, the better the voice assistant’s responses will be as it continues to “learn” from the data it gathers. And for call analytics and customer data acquisition, thirty words spoken is much more valuable than a single word response.
Beyond the ability for callers to speak naturally, AI-powered voice assistants are also becoming more empathetic and able to handle emotionally sensitive interactions appropriately and sensitively. For instance, in customer service interactions involving emotionally charged topics like bereavement, AI voice assistants can adjust their tone and communication style to be more empathetic. This can manifest in a softer speaking voice, the use of more compassionate language and a slower, more patient conversational pace. Some customers find interacting with a sensitive and understanding voice assistant less emotionally taxing compared to a potentially awkward human interaction.
Research has shown that customers perceive AI assistants as more patient, especially when dealing with repetitive inquiries. Additionally, AI excels at delivering quick and consistent answers to straightforward questions. These factors make AI chatbots well-suited for handling certain basic customer service tasks, offering the contact center a valuable tool to streamline operations and take some strain off their human agents.
Ethical AI
AI disclosures are a slippery slope. Some businesses choose upfront transparency, informing users from the outset that they are interacting with a virtual assistant. The idea here is that transparency fosters trust and allows users to manage their expectations accordingly. On the other hand, some companies employ a more subtle approach, seamlessly integrating AI into the interaction without explicitly mentioning its presence. This strategy aims to prioritize a smooth user experience and avoid reminding users of negative perceptions about voice AI’s capabilities. Because honestly, if you announce upfront the caller is speaking to a bot, they’ll be less likely to engage naturally for all the reasons mentioned above.
Legislation is still evolving in this area, and future regulations may mandate specific disclosure requirements for companies leveraging AI in customer service interactions. However, the current focus for many companies seems to be on designing AI experiences that bridge the gap between efficiency and user satisfaction. This involves leveraging virtual assistants with advanced conversational abilities that can not only handle routine tasks but also adapt to complex inquiries, fostering a sense of natural interaction without necessarily revealing the technology behind it. No one is trying to “trick” users into believing that the voice assistant they’re talking to is human; rather, it’s about creating AI that is so helpful, it doesn’t matter if it’s human or not, and the caller is likely to ignore the fact they’re talking to machine.
AI systems are becoming increasingly advanced and human-like, so it’s critical for regulatory leaders to balance innovation with ethics. There must be space for organizations to continue to enhance their AI’s abilities for customer experience, but business ethics must also remain front and center in conversations about AI.
As AI continues to mature and become more sophisticated, the line between human and AI interaction will continue to blur. Businesses that embrace this evolution and strategically integrate AI assistants into their contact center operations will be well-positioned to elevate customer experience. AI's ability to handle routine inquiries efficiently and consistently will free up human agents to focus on complex issues and provide personalized support.
However, achieving this balance requires careful consideration and a focus on building trust with customers and managing expectations. AI voice assistants should be designed with emotional intelligence in mind, allowing them to recognize and respond to customer sentiment appropriately. Additionally, ethical considerations around the responsible use of AI will be paramount. By prioritizing these aspects, businesses and customers can unlock the true potential of AI-powered customer service, fostering a future where automated efficiency seamlessly blends with empathy and personalized care to create excellent customer experiences that drive business success.
###
About the Author
Nikola Mrkšić is the Co-founder and CEO of PolyAI. He graduated from the University of Cambridge with a Ph.D. in Machine Learning and Natural Language Processing.
Nikola was named in the Forbes 30 under 30 in 2021 for his work with PolyAI.