Contrary what some think, AI is not new. It’s a function of computing that has been around for a long time. What’s new is the growing acceptance of AI for expanded applications and its ability to transform data in actionable business strategy.
Daniel Ziv, GTM Strategy VP, Experience Management and Analytics, noted that AI has been around a long while. Verint. AI is not a single thing. It can have a wide range of capabilities, depending on the task it was designed to perform.
Large language models (LLMs) are a well-established component of AI. Around 18 months ago, advances in natural language generation and understanding led to a significant increase in the capabilities of these systems.
Ziv told E-Commerce Times that “that work has evolved and built for many years.” “It was a great way to raise awareness, because anyone could try it.”
AI’s pivotal shift in business
Ziv said that a meaningful turning point has accelerated both the need for and the opportunities for automation platforms, which organizations can use in new ways. As an example, generative AI continues to evolve and become smarter at understanding languages.
Cloud computing is one of the key elements in AI adoption by businesses. It allows for more data to be processed faster, and at a cheaper cost. In the past ten years, most AI software was deployed on-premises. Adopters were required to purchase hardware, configure it, install the software, and train everyone.
“It would take months — sometimes years — to get the value that now you can get sometimes in days or weeks,” Ziv said.
Today, the challenge is to understand how AI has advanced in the past two years and leverage it to transform large data sets for analysis and recommendations. There are many ways to transform data, depending on whether it is structured or unstructured.
“Structured Data tends to be Numbers, and Computers have been running structured data.” He said that computers are good at creating models and performing tasks based on numerical data.
Unstructured or semi-structured data is more difficult to transform, as it includes elements such as text, audio and video, and metadata.
“In the old days, this was more difficult for computers. Ziv explained that with the advent of generative AI technology, computers can now do this much more quickly.
Refine AI for Tailored Business Intelligence
Verint uses AI to help businesses use their data more efficiently. It has assisted its customers in addressing a wide range of accuracy problems.
“In our industry I think people may perceive transformative data as not being accurate, because we’ve trained general LLMs on internet data that does not pertain to your business. It’s not behavior data. Ziv said that it was similar to a baby learning to talk.
Our AI has been trained to understand and respond to language on a general level. He said that the AI’s ability to understand is like a child who does not have the necessary knowledge, information and experience to answer questions in a way that is relevant to the intended results.
AI developers are still learning how to grow that baby into an adult. Ziv believes that the best solution is to combine this ability to comprehend language with behavioral data to create language specific to interactions between you and your customers, or organisations with their customers.
“We’re just beginning this transformational phase.” “I do believe, however, that the ability to write data on an open platform with the power of generative artificial intelligence will enable us to see compelling things and allow us automate,” he said.
The Journey to Predictive Accuracy
SoundCommerce Here’s an example that shows why data-driven predictions of actionable outcomes are not a universal process. The company uses a no-code platform that is accessible to all.
Eric Best noted that the data-transformation pathway is a minefield of challenges. The data transformation process includes extracting customer data and data from an external source system.
After that, it is necessary to validate the data to make sure they are of reasonable quality. Best said that the next stage is to apply the data to a problem SoundCommerce wants to solve. That is, to assign meaning to the data flowing.
Best told E-Commerce Times, “This is important because you’re going to be making these critical business decisions by the time that you reach the warehouse, where the analysis will take place.”
For this to happen, data must be converted to create compatibility. For most retail brands orders come in from different sources, not just a cashier or a point-of sale system. These vending sites include an Amazon Marketplace storefront, e-commerce, and proprietary mobile apps.
Best explained that AI could be useful in this area, as it would allow all four order records to be presented with a standard format and schema.
AI Mapping without an Engineering degree
In order to achieve accurate results, you must be able define the data in terms of natural language. In order to have the AI help you with this problem of data mapping, you must tell it in very detailed natural language what you are looking for and how you define that data.
Best explained that the solution to this problem is to have the AI create the software that will change the data. Best explained that you don’t need to be a great software engineer but rather a prompt engineer.
People must be very good in explaining what they want in natural language, and not in code terms. It is important to be accurate in both speech and writing.
SoundCommerce’s customers are only just starting to experiment with these AI algorithms. Best explained that the company uses its proprietary algorithms to enable AI in some cases.
One example of a proprietary code is the ability forecast the lifetime value for an individual customer. Microsoft, Google and Amazon Web Services are the innovators in generative AI capabilities. Snowflake Best’s works is an example of a successful company.
Cloud platform companies generally build their own AI tools with large language models.
Modern AI and Timeless Business Questions
How cost-effective is AI for decision-making in business? Best said that the answer depends on what you are asking.
The more you can narrow down a specific use case, the more practical the new technology becomes for less-technical organizations. SoundCommerce learned this lesson the hard way.
Best answers the question of cost-effectiveness versus practicality using an ancient truth. Since more than a hundred years, people have tried to figure out how to spend their advertising dollars efficiently.
The questions and answers aren’t new. “The ability to automate answers at scale is certainly new”, Best concluded.
