Red paper plane out of line with white papers to represent change and disruption.
Editorial

ChatGPT and Customer Service: Get Ready for an Epic Disruption

4 minute read
Muddu Sudhakar avatar
By
SAVED
ChatGPT’s core technology — generative AI — is poised to transform just about all industries.

The Gist

  • Game-changer for customer service: OpenAI's ChatGPT rapidly surpasses 100 million users, demonstrating the mainstream adoption of generative AI, especially in transforming the customer service industry.
  • Evolving into next-level customer support. Generative AI marks a significant milestone in customer service technology evolution, greatly improving performance and handling complex tasks within seconds.
  • Not all smooth sailing. Despite its benefits, generative AI implementation presents challenges like "hallucinations," lack of explainability, and privacy concerns, requiring robust, secure, and accurate platforms for effective use.

Not many apps have reached 100 million users. Usually, this feat takes years. Facebook hit this milestone in 5+ years and Snap did it in 3+ years.

Then ChatGPT came along like a hurricane. It crossed over 100 million users in about two months. And it seems like the momentum is just starting.

ChatGPT has instantly made AI mainstream. It is educating users on how to use natural language prompts to get human-like responses, which represents a new paradigm in technology. 

ChatGPT’s core technology — which is called generative AI – is poised to transform just about all industries. But there are some that are likely to feel the impact sooner. Just look at customer service. Already, a variety of the software vendors have retooled their systems with generative AI functions.

This technology will mean more engaging and effective interactions with customers — done at scale. There will also be significant cost savings. 

But navigating this will not be easy. Generative AI needs to be approached carefully to get the best results. 

Evolution of Customer Support Technologies

Until the 1970s, customer support was mostly about agents answering phones. If there was automation, it was about accessing enormous mainframes systems.

But the PC revolution would dramatically disrupt the industry. What emerged were IVR systems that could process input from callers. This would eventually lead to voice recognition.

Then AI became a major factor during the past decade. This resulted in the widespread adoption of chatbots. Business certainly realized benefits from these technologies. It was easier to track service. There were real cost savings. And customer service levels improved.

Despite all this, these technologies were not particularly smart. They were essentially hard-wired solutions for predefined scenarios, with a bunch of IF/THEN logic. This meant that it was difficult to handle more complex calls.  After all, a common action for callers has been to keep pressing the “0” button to get the attention of a human agent.

But generative AI represents a quantum leap in capability. There is now real intelligence. A bot can understand open-ended prompts and create engaging responses. 

In terms of the benefits, they will be much more impactful than prior innovations. While IVR and chatbots may have seen 5% to 10% improvements, generative AI is poised to show levels of 50X or more.

Related Article: Generative AI Solutions for the Contact Center

The New Game of Generative AI

Generative AI should resolve most tier-1 support, which is geared for solving issues like resetting passwords. Granted, traditional chatbots can do much of this already. But generative AI will mean that the experience will be like interacting with a human. This should result in higher satisfaction.

But the huge change will be for tier-2 support, which is a large expense item. This often involves a human agent that has to work across multiple IT systems, reading through long and complex knowledge-base articles and writing responses. The process can be time intensive. It could also require the help of other human agents.

With generative AI, a bot can work across any system and summarize massive repositories. It can also write compelling content, say in a certain tone that is appropriate for the scenario. All this can be handled in seconds.

If the generative AI cannot handle a particular question, it can quickly route the issue to the right human agent. The system can then learn from this — which means a similar situation will be automated for the future. In other words, there will be a flywheel that gets better and better.

Related Article: Real Marketing and Customer Experience Questions — and ChatGPT's Answers

Generative AI Comes With Customer Support Challenges

With generative AI, enterprises will need a radical new approach to operating customer service. Perhaps a key role will be to have AI systems managers. They will be the human-in-the-loop to ensure the smooth operation of the system. 

They will also help deal with the complex issues of generative AI. For example, the models can suffer from hallucinations. This is when the answers are convincing but false or misleading.

Of course, when dealing with customers, accuracy is essential. Otherwise, there will be a clear threat to an enterprise's brand. 

Another problem with generative AI is that the model is a black box. There’s no way to understand how the system has spun up the responses. After all, these models are massive and mind-numbingly complex. 

But for many enterprises — especially those that are heavily regulated — there needs to be a basis for information. It cannot be from the whimsy of a highly sophisticated algorithm. 

Then there are the issues with security and privacy. Keep in mind that customer information will be processed remotely via an API. This can be a dicey proposition.

In light of all this, it’s important to rely on a platform that is built for the rigorous needs of enterprises. It needs to strive for high levels of model accuracy, explainability and security. 

Learning Opportunities

Conclusion: A Generational Change for Customer Service Technology

Generative AI represents a generational change in technology. Unlike traditional systems, the responses from the models will be like communicating with a human. But there will be the benefits of processing large amounts of complex data in little time. 

Yet there must be an understanding of the inherent risks of generative AI. If not, they can easily result in damage to an enterprise’s brand if not managed properly. This is why it’s important to focus on a system that is enterprise grade.

fa-solid fa-hand-paper Learn how you can join our contributor community.

About the Author
Muddu Sudhakar

Muddu Sudhakar is a Silicon Valley based entrepreneur, and currently the CEO of Aisera. Having built several applied AI companies that have been bought by some of the most celebrated companies in Silicon Valley (ServiceNow and Splunk) he is considered a leading voice in AI/Machine Learning, IoT, Big Data, and CX. Connect with Muddu Sudhakar:

Main image: Dilok
Featured Research