Generative artificial intelligence (AI) is able to combine and analyze data from various sources in order to create more targeted customer segments than traditional customer data analysis. Generative AI is also able to discover distinct patterns in customer behavior, enabling brands to create more useful rules for each segment and marketing campaigns that are more suitable for those segments.
Additionally, generative AI can provide actionable insights on emerging trends for specific customer segments over time, allowing brands to enhance and improve their strategies to reach their audience segments more effectively. Let's take a look at generative AI, its uses and capabilities, along with the challenges of using generative AI for customer data analysis.
Artificial intelligence is already used by customer data analysts to clean, analyze, explain and visualize customer data. The idea of using generative AI for customer data analysis is a new frontier, given that generative AI is still a relatively new and evolving technology.
There are many generative AI applications that are being developed by tech industry leaders, including OpenAI’s ChatGPT, Google’s Bard, Microsoft’s Bing and Level AI, which just announced its AI system for customer service teams, AgentGPT. Currently, the only standalone generative AI application that is publicly available to try out is ChatGPT (generative pre-trained transformer), a large language model which was created by OpenAI. Through a combination of GPT-1, GPT-2 and GPT-3, it was trained on a massive amount (over 45 terabytes) of data from the internet, including books, articles, websites and other sources up until 2021. ChatGPT is able to perform a wide range of natural language processing (NLP) tasks including text generation, summarization, translation and sentiment analysis.
Microsoft’s new AI-driven Bing, created in partnership with OpenAI, is based on a newer version of GPT, was trained on data that is current, and is able to search the internet for more up-to-date, real-time data. It, along with Google Bard, is still being tested by a limited number of users. Recent announcements from both companies were soon followed by news stories denouncing the generative AI models for providing incorrect information, as well as humanlike biases, negative traits and even seemingly sentient anger. Although there is still plenty of work to be done before these generative AI applications are ready for prime time, once ready, there are many ways they can be used effectively for customer data analysis.
Although much of this discussion is still speculative, generative AI promises to be able to be used for customer data analysis for the following applications:
By analyzing large amounts of data and identifying patterns and trends, generative AI can also help brands to obtain actionable, data-driven insights, which can be used to improve interactions with customers and the overall customer experience.
Matt Hallett, head of product solutions at Amperity, an enterprise customer data platform, told CMSWire that generative AI has the potential to revolutionize how companies operate through automatic processes and tasks and even generate original content by continually learning from existing data. "Generative AI could provide a well-packaged foundation to understand customer behaviors and preferences, a strong foundation that companies can build upon, but it might not have the creativity — that human touch needed to help differentiate a brand." Additionally, Hallett raised a valid concern, suggesting that if everyone is using generative AI, companies risk their messaging sounding like everyone else.
Related Article: Level AI Debuts Generative AI Tech for Contact Centers: AgentGPT
In the near future, generative AI applications are unlikely to have the same level of understanding and context as human data analysts. Customer data analysts are educated and experienced with skills and expertise in data analysis that surpass that of current generative AI models.
“Generative AI takes the guesswork out of what customers want, freeing up time and resources spent on solving similar problems over and over again,” said Shaunak Amin, co-founder and CEO at SwagMagic, a global branded swag producer, retailer and distributor. “This way, data analysts and support teams can work together to find ways to create highly personalized, instant and effortless experiences for individual clients.” Amin believes that these types of experiences help to create loyal customers. “For this reason, it's vital that employees consider generative AI as a helpful tool rather than a threat to their jobs.”
Human analysts know which questions to ask, and although AI is able to identify patterns and trends, humans are able to create hypotheses that can be tested using additional analysis. By combining these skills with knowledge and insights from other areas of a business, human analysts are able to provide a more holistic, 360-degree picture of the customer.
“While generative AI has the potential to provide useful insights into customer behavior and emerging trends, there are certain things that a human can do better than a machine,” said Sinoun Chea, digital marketing consultant and CEO of ShiftWeb, a digital marketing and web design company. “For example, analyzing customer feedback and conducting research on target market segments requires a more detailed understanding of people's needs and behaviors, as well as complex problem-solving skills that only humans possess.” Chea explained that customer data analysts have the ability to think abstractly and outside of the box in order to identify customer needs and tailor marketing messages accordingly.
Related Article: Zuckerberg Announces New Top-Level Team Focused on Generative AI
The quality of the content that generative AI produces is directly proportional to the data that it is trained with. The larger the data set and the better the data, the better the results will be, and for customer data analysis, this is particularly applicable. Low-quality data, biased data or limited data will result in inaccurate insights and conclusions being drawn from the generated data.
Given that generative AI will require large amounts of customer data in order to be effective, there are understandably data privacy concerns about using a generative AI model to analyze customer data. Some data, such as customer feedback and survey results, would not pose any privacy risks, but detailed customer order history, personal details and other demographics may be likely to generate data that contains sensitive personal information.
Microsoft’s new Bing has recently been in the news for specifically naming a reporter that it “felt” had injured it by publishing what it deemed to be “negative feedback” about the search engine, referring to the reporter as “ugly” and comparing them to Hitler. At this point in time, generative AI is still largely unpredictable, and is likely to reflect the biases of those who originally created the data that it has been trained on. If the training data contains bias against a specific demographic, the data it generates may also be biased against that demographic.
“Along with the positive aspects, there are also pitfalls that require business leaders to proceed cautiously,” said Hallett. “The technology presented today has many ethical and practical challenges. For example, natural language processing (NLP) models can present false information based on rephrasing or summarizing the original content incorrectly.”
Thus far, generative AI is not based on explainable AI (XAI), so the data that it generates originates from what is considered to be a “black box” in that it is difficult to understand how and why it generates specific results. When generative AI is used to analyze customer data, it would be challenging to explain why it came to the conclusion it did about a specific customer segment, for instance.
OpenAI allows access to GPT-3 through an application programming interface (API), so developers can use its generative AI chat in their own applications. Yext AI recently announced that its generative AI chatbot will be available for businesses, a trend that is likely to continue to evolve and grow in the future. However, because generative AI needs to be trained on a huge data set, the computational resources that are needed can be expensive and time-consuming, and largely out of reach for most brands. Integration with existing customer data analysis systems is also largely challenging if they were not designed to work with generative AI.
Generative AI has the capacity to be used for many different applications across multiple industries. The technology shows promise as an effective tool for customer data and predictive analytics, but rather than replace human data analysts, generative AI models will be used to enhance and improve their work, making it less costly and more efficient.
