ICMI is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Advertisement

I Ask ChatGPT If It Can Improve Contact Center Customer Experience

robotThere’s been so much talk about how ChatGPT could replace contact center jobs that I thought I’d just ask ChatGPT about it. I specifically asked about the accuracy of information (this has been a well-documented issue with Large Language Models), policy exceptions, and data security.

 

Rob Dwyer: Can ChatGPT be effectively incorporated into contact centers to improve customer experience?

ChatGPT: Yes, ChatGPT can be effectively incorporated into contact centers to improve customer experience. Here are some ways in which ChatGPT can help:

  • Instant Response: ChatGPT can provide instant responses to customers' inquiries, reducing the waiting time and improving the customer experience.
  • 24/7 Availability: ChatGPT can work 24/7, providing assistance to customers even outside working hours, which can lead to higher customer satisfaction rates.
  • Personalization: ChatGPT can personalize interactions with customers by using their name and history to provide relevant information and recommendations.
  • Multilingual Support: ChatGPT can support multiple languages, which can be useful for contact centers dealing with customers from different regions.
  • Scalability: ChatGPT can handle an unlimited number of conversations at the same time, which can improve the efficiency of the contact center and reduce costs.
  • Analytics: ChatGPT can provide valuable insights into customer behavior, which can help contact centers optimize their processes and improve the overall customer experience.
    Overall, incorporating ChatGPT into a contact center can lead to a more efficient and effective customer service experience, which can improve customer satisfaction and loyalty.

Rob Dwyer: How would ChatGPT handle customer inquiries when there is conflicting information about how to resolve their issue?

ChatGPT: When there is conflicting information about how to resolve a customer's issue, ChatGPT would use its natural language processing and machine learning algorithms to identify and understand the nature of the conflicting information.

Then, ChatGPT can provide the customer with different options and explanations for the conflicting information. ChatGPT can also ask the customer more questions to better understand their specific situation and provide a more personalized solution.

If ChatGPT is unable to resolve the issue, it can transfer the conversation to a human customer service representative who can provide further assistance.

Overall, ChatGPT can help mitigate the impact of conflicting information by providing customers with various options and explanations and gathering more information about their specific situation, ultimately leading to a more satisfactory resolution of the issue.

Rob Dwyer: What if human customer service representatives are using ChatGPT to resolve customer issues and there is conflicting information?

ChatGPT: If human customer service representatives are using ChatGPT to resolve customer issues and there is conflicting information, the representatives can follow these steps:

  • Identify the source of the conflicting information: The representatives can identify where the conflicting information is coming from, whether it's from the customer, the company's documentation, or other sources.
  • Verify the information: The representatives can verify the information by checking multiple sources and confirming which information is correct.
  • Communicate with the customer: The representatives can communicate with the customer and explain the conflicting information and what the correct information is. They can also apologize for any confusion or inconvenience caused by the conflicting information.
  • Resolve the issue: Based on the correct information, the representatives can work with the customer to resolve the issue in the most appropriate way.
  • Update the system: The representatives can update the company's documentation or systems to ensure that the correct information is used in the future, reducing the likelihood of conflicting information arising again.

In situations where conflicting information arises, it's important for the representatives to communicate with the customer and work collaboratively to resolve the issue, ultimately leading to a positive customer experience. ChatGPT can assist the representatives by providing relevant information and helping to identify potential solutions, but the human touch is still critical in these situations to ensure the customer's needs are fully met.

Rob Dwyer: Does ChatGPT provide the sources of information it provides, particularly when there is conflicting information, so that representatives can verify the sources?

ChatGPT: ChatGPT can provide sources of information it provides, particularly when there is conflicting information, so that representatives can verify the sources.

ChatGPT can be programmed to provide sources of information as part of its responses. For example, it could say "According to our product documentation, the correct way to resolve this issue is...".

If ChatGPT provides conflicting information, it can provide sources for both pieces of information and ask the customer or representative which source they would like to follow.

ChatGPT can also be designed to provide more detailed information on sources upon request. For example, it could say "Would you like me to provide a link to our product documentation that supports this information?"

Providing sources of information is important for transparency and credibility, and it can help representatives verify information and ensure that they are providing accurate information to customers.

Rob Dwyer: Could companies rely on ChatGPT to make good decisions about when to make exceptions to policies for customers?

ChatGPT: While ChatGPT can be programmed to follow company policies and guidelines, it is not recommended to rely solely on ChatGPT to make decisions about when to make exceptions to policies for customers.

ChatGPT is designed to provide responses based on the data it has been trained on and the rules it has been programmed with. However, making exceptions to policies often requires human judgment, empathy, and an understanding of the specific situation and context.

Human representatives may be better equipped to evaluate the customer's circumstances and make exceptions to policies when necessary. They can consider various factors, such as the customer's history with the company, the severity of the issue, and the impact on the customer, among others.

ChatGPT can assist representatives in making decisions by providing relevant information and options, but the ultimate decision-making responsibility should be with the human representative.

In summary, while ChatGPT can provide valuable insights and support for decision-making, it should not be relied upon solely to make decisions about when to make exceptions to policies for customers. Human judgment and empathy are crucial for ensuring that decisions are made in the best interest of the customer and the company.

Rob Dwyer: Is it possible for ChatGPT to allow access to private customer information by accident if it has access to customer data?

ChatGPT: ChatGPT can potentially access and share private customer information if it has access to such data. However, there are several measures that can be taken to minimize the risk of this happening.

  • Firstly, access to private customer information can be restricted to authorized personnel only. ChatGPT's access to customer data can also be limited to specific data points and information deemed safe for sharing. Additionally, ChatGPT can be programmed with data security protocols to prevent it from sharing any customer information without proper authorization or encryption.
  • Secondly, ChatGPT's data processing and storage can be encrypted and secured with advanced security protocols to minimize the risk of unauthorized access. ChatGPT can also be designed with audit trails to track and monitor any access to customer information.
  • Thirdly, regular security checks and penetration testing can be conducted to identify and address any vulnerabilities in the system.

Overall, while there is a potential risk for ChatGPT to access and share private customer information, proper measures can be implemented to mitigate this risk and ensure that customer data is kept safe and secure.

My Conclusions

While ChatGPT presents lots of intriguing opportunities related to contact centers, thoughtful leaders will consider the ramifications of relying on AI to provide information to customers and how much information it should have access to if they move forward.

While ChatGPT said it could provide sources for conflicting information, it’s been known to not only provide false information, but to also create false sources to back up that information.
Allowing it access to customer data presents a self-admitted risk which should be taken seriously given its previous issues in protecting private data. Imagine the PR nightmare and potential financial ramifications of a customer data breach that involved simply chatting with your custom ChatGPT bot.

Finally, even ChatGPT recognizes that humans are “better equipped to evaluate the customer's circumstances and make exceptions to policies when necessary.” So while contact center leaders may augment their operations with AI, the human contact center agent isn’t going anywhere any time soon.