All Posts

ChatGPT in the Lens of Customer Experience…What’s Next?

April 19, 2023
ChatGPT in the Lens of Customer Experience…What’s Next? featured image

Understanding ChatGPT’s abilities, restrictions, and applications in customer service is essential to comprehend its potential impact on the future of CX.

Organizations worldwide are in constant pursuit of enhancing customer experience (CX), and several companies are increasingly relying on cutting-edge Artificial Intelligence (AI) language tools to achieve this goal. Among the latest AI-driven conversational agents making headlines is ChatGPT, launched by OpenAI in November last year. This interactive chatbot harnesses machine learning technology to provide sophisticated and hyper-personalized solutions to users.

While chatbot technology has undergone continuous improvements over the last decade, the release of ChatGPT, a Large Language Model (LLM)-based chatbot, has heralded a new era of highly adaptable and powerful conversational agents. Developed on OpenAI’s GPT-3 family of LLMs, ChatGPT represents a significant breakthrough in natural language processing (NLP) by possessing the ability to comprehend intricate natural language queries. The software and other AI-based models are essentially a step up from the technologies many contact centers have used to support their customer support transactions for years.

The allure of ChatGPT’s capabilities has caught the attention of businesses across multiple industries. Already, AI has demonstrated its potential in improving productivity, fostering creativity, and enhancing daily operations for both individuals and organizations. Therefore, it comes as no surprise that companies are keen to leverage ChatGPT’s capabilities in revolutionizing customer service. On the other hand, there are also criticisms and risks of ChatGPT in the contact center.

This article will examine what is possible and what is not with ChatGPT. We’ll also dive into the potential applications in customer support that are beyond the hype, and the technology’s probable impact on the future of CX.


What makes ChatGPT special lies in its explicit optimization for facilitating conversations. In sharp contrast to prior iterations of LLMs, ChatGPT exhibits the ability to retain knowledge garnered from previous exchanges, thereby enabling the capacity for seeking clarification and questioning incorrect responses.

Some brands are already figuring out how to harness the technology to improve online chat functions. Meta, Canva, and Shopify, among other companies, are already using the technology in their customer service chatbots. Leveraging the capabilities of LLMs, chatbots have developed amazing competencies to generate human-like responses and to speak in different languages and styles. According to Juniper Research, it’s predicted that AI-powered chatbots will handle up to 70% of customer conversations by the end of 2023.

This highlights the growing reliance on AI to enhance CX and streamline interactions to tackle customer service communication that is not straightforward and can benefit from a conversational, powerful, and intelligent bot. As an AI language model, ChatGPT’s capabilities in customer service and customer support are quite extensive.


Faced with these new technological possibilities, we see CX leaders questioning how to take advantage of this new technology to reimagine the digital customer experience.

Generative AI tools like ChatGPT have the potential to be transformational — likely impacting every aspect of business, especially customer support.

Here are just a few areas where ChatGPT could influence CX.

Automated Customer Service
ChatGPT can be trained to identify common customer inquiries and issues such as shipping updates, billing inquiries, and product information. By doing so, it can provide quick responses to customers without the need for human intervention. This can save customers time and reduce the workload of customer service agents.

24/7 Availability
Unlike human customer service agents, ChatGPT is available around the clock. Customers can receive assistance anytime they need it, which can improve customer satisfaction and loyalty.

Personalized Support
By using customer data, ChatGPT can offer personalized support. For example, if a customer has made previous purchases, ChatGPT can suggest similar products or provide discounts based on their buying history. This level of personalization can make customers feel valued.

Multilingual Support
ChatGPT can be trained in multiple languages, enabling brands to offer support to customers globally. This can help brands expand their customer base and improve satisfaction for non-native speakers. ChatGPT can provide a more natural and personalized experience for customers by communicating in their language.


One potential use case for advanced language models like ChatGPT is in providing subject matter expertise. With the ability to process large amounts of unstructured data, these models can generate coherent responses to semi-structured natural language queries, enabling chatbots and IVRs to provide tailored responses to customers.

Another potential application is in agent assist, where the underlying NLP algorithms can aid agents in their interactions with customers. Language models can help agents understand customer intent and suggest relevant response options and knowledge articles based on real-time conversation analysis. Language models can also be used for intelligent routing, leveraging their ability to understand customer needs to match them with an agent with the appropriate skill set. This provides a more accurate and efficient routing process, with the added benefit of providing agents with a summary of the customer’s needs.


Despite the hype, CX leaders may be wary of AI’s tendency to get things wrong. The model is strictly based on its trained parameters and the previous words in a conversation. It cannot access any outside information or database to make its predictions, and it doesn’t know anything that happened after 2021.

In speaking with Matt Taylor, Chief Product Officer at Knowbl, we discussed the critical functionalities that could mean the difference between delivering great CX at scale and a frustrated customer. Knowbl is an enterprise-ready virtual concierge platform that leverages on-brand, compliant content.

“The one thing to keep in mind is that Open AI is providing an API and not a platform,” says Taylor. “Very few companies have the resources internally to be successful with taking that API and building the necessary functionality around it to be successful.”

With that said, an overreliance on ChatGPT could lead to brands sharing incorrect information with customers without realizing it, which is why human judgment needs to be applied to avoid errors or bias.

“ChatGPT is not safe for direct customer service or support (yet),” adds Taylor. “It is extremely helpful for generating content that can be used for customer support such as FAQs, blogs, or articles but there has to be a layer of human review and approval before anything can be customer-facing. The reasoning for this is that Generative AI tends to not only answer questions inaccurately but, in some cases, will completely fabricate an answer that has absolutely no truth or substance behind it. This has commonly been referred to as ‘hallucinating.’ But the AI is not hallucinating. It is using the large mass of data it has been trained on to calculate/generate the most likely token (e.g., word) after the previous token. And these models are black boxes, meaning there is no way to explain why the model decided to answer the way that it did. With no explanation, there is no clear path to successfully tune the model.”

Moreover, ChatGPT models may not be capable of adding value to customer queries that are repetitive in nature and require consistent answers. Some of the most asked questions might be, “Where’s my package?” “How long will my return and refund take to process?” or “What’s my password?”

Whereas many chatbots are trained to deliver a response of, “I don’t know” to requests they cannot handle, ChatGPT, for example, is more likely to respond with complete confidence — even if the information is incorrect.

“Knowbl leverages the underlying technology of LLMs, but has also created a safe and compliant application for the enterprise by leveraging their content to power the AI. Our model will not “hallucinate” because it will not generate its own response,” explains Taylor. “It will only pull from the content the AI has ingested from the brand. If content does not exist for the user question, it will not be able to answer, as opposed to generating a completely fabricated response that has no basis. This is what allows us to be directly customer-facing when it comes to service and support.”


As an AI language model, ChatGPT can work in conjunction with human customer service agents as a productivity-enhancing tool as opposed to a complete replacement for customer support and customer service. According to Forrester, the short-term achievements will likely outweigh the long-term success for organizations that are investing in AI as a solution to simply cut costs and automate tasks in the process.

Here are just a few of ChatGPT’s current limitations.

Limited ability to empathize: While it can provide helpful responses to customer inquiries, it lacks the emotional intelligence and empathy of a human customer service agent. As such, it may not be able to fully understand or address the emotional needs of a customer.

Difficulty with complex issues: Although it can handle a wide range of customer inquiries, there may be some complex issues that it’s not equipped to handle. In these cases, it may be necessary to transfer the customer to a human agent who has the expertise to address the issue.

Inability to handle non-textual communication: As a language model, it can only process and respond to text-based communication. It’s unable to handle voice calls or video chats, which may be necessary for certain types of customer support interactions.

Limited knowledge of certain specific domains: ChatGPT’s knowledge base is extensive but not exhaustive, and there may be specific domains or industries where it lacks the required expertise to provide accurate and helpful support.

Clearly, ChatGPT still has many shortcomings (e.g., hallucinations, biases, and non-transparency), but from the looks of it, the technology is quickly advancing.


In conclusion, there is no question we will see a transformational shift in CX over the next few months and years, with AI and ChatGPT tools driving accelerated change. What we’re seeing is that ChatGPT can serve as a tool for brands looking to improve their customer service and support. By combining the strengths of ChatGPT with human customer service to provide automated responses, 24/7 availability, personalized support, and multilingual support, ChatGPT may help brands meet the demands of today’s customers and improve their customer satisfaction and loyalty.

However, empathy is critical in any type of CX engagement. Therefore, ChatGPT in its current form could never replace customer service agents. Despite advanced sentiment analysis and NLP, there are some circumstances — even transactional ones — where human connection is the best strategy for long-term customer loyalty.

These tools are profoundly changing the way we work, and how organizations operate. We’re excited to see how brands will continue harnessing the power of AI models to drive transformation.