Guiding the Customer Journey With Generative AI
ChatGPT has dominated headlines across all industries, and while it may be tempting to just jump in, it’s important for retailers to take note of the positives and negatives that come with this new tech tool; a new form of generative AI.
While ChatGPT may be in its earliest stages, the possibilities for generative AI to become an essential component of customer service are already apparent. When it comes to customer service and the customer journey, there are three important facets: privacy, precision, and emotion. Let’s take a closer look at how generative AI plays into each.
Trust is key in the customer journey, and when it comes to privacy, public generative AI tools like ChatGPT use inputs to learn and then train their models, so users and organizations alike should be wary of what information they input in the system. However, ChatGPT does also only draw from the public domain, and excludes critical customer information such as medical and financial data. Privacy will likely continue to be a closely monitored facet of generative AI.
Another important element of customer service and consumer trust is precision. ChatGPT will need to advance accuracy before companies or consumers can fully rely on it. Let’s say humans have about a 98% accuracy rate (accounting for human error), while most well-tuned generative AI tools fall in the 80-90% range. When it comes to customer service, being 100% correct is imperative, especially with private or sensitive information, so there will still be a need for human intervention and fact-checking.
Despite its struggles with accuracy within consumer use, generative AI can unlock greater insights for brands to improve business results, time to value, and the employee experience. For example, generative AI can be used to cluster data into more effective segments for targeted customer experience, or to manufacture "dummy data" to test and refine AI models for conversion or retention.
Human intervention is essential in fact-checking ChatGPT, and it also comes into play with gut-checking emotion in generated responses. Surprisingly, human emotion may be easier to generate than you think. ChatGPT uses natural language processing (NLP) to facilitate “plain English'' conversations between users and provide answers in less than five seconds. NLP has made large strides toward the raw emotion humans have when delivering customer support. Though many people still think bots could never replace a human, the more the bot learns, the more human-like it will seem.
While humans are currently the only writers capable of true empathy, NLP has made great improvements with emotion detection. All of this relies on the quality of training data, architecture choice, and algorithm tuning — things that are facilitated by humans.
As generative AI’s popularity is only set to increase, we will continue to see retail and other industries attempt to automate to continually advance and scale the customer experience. As long as privacy, precision, and emotion are at the forefront, generative AI can be a powerful tool to deliver extraordinary customer experiences.
— Eric Carrasquilla, President of Customer Experience,CSG
