The term “AI” actually refers to any machine system that attempts to perform a task intelligently. This means trying to use what you know or can find to perform your task more efficiently, rather than just blindly following a set of instructions.
The most common way to achieve this in business today is through a process known as machine learning (ML). This refers to data-trained algorithms that make decisions based on what they know and improve as they learn.
Machine learning algorithms can be further classified into several specific technologies. These include computer vision (allowing computers to see and understand visual data), natural language processing (computers that can understand and communicate with human language), and recommendation engines (which Amazon, Spotify, and Netflix use to predict what customers might want to buy or consume).
AI business application
Applications are specific use cases for AI within a business, and they typically fall into one of many high-level categories. This is what I will explain today.
Before we start, please note that there is a lot of duplication. A given AI implementation may fit more than one of these categories. As time goes on, things like this can become even more common.
To understand the differences and explore some use cases in these categories, Davor Bonaci, CTO and Executive Vice President of DataStax, joined my podcast. Bonaci has been with DataStax since earlier this year when he acquired Kaskada, his ML business he co-founded. Prior to that, he worked as an engineer at Google Cloud.
Categories of AI Applications
So let’s start with last year’s hot topic. It’s definitely generative AI. From ChatGPT, the fastest growing application of all time, to more niche and experimental tools focused on creating videos, music, or 3D designs. Generative AI applies to any application that uses AI (ML) to create something. This could mean personalized products that use AI to input information to create customized or bespoke customer-facing products. It can also mean everything from computer code, marketing materials, product designs, responding to customer service requests (chatbots), to synthetic datasets used to train other ML models and build digital simulations.
Mr. Bonaci tells me: “Everybody talks about generative AI.
“We have found this to be very productive. This is a very interesting innovation and everyone is looking at it.”
Similar to generative AI, an important AI category for businesses is predictive AI. The theory behind ML has been around for the better part of his century, but it’s only in the last decade or so that cost and accessibility (in terms of computing power) have reached a level where it’s an option for nearly every business. Predictive AI helps businesses predict the future, including sales forecasting, customer lifetime value estimation, customer churn prediction, risk assessment, fraud detection, marketing collateral distribution targeting, predictive maintenance of machinery, and supply chain logistics optimization.
“Models that predict the future based on what happened in the past will help companies enable and predict customer behavior, anticipate market demand, optimize operations, and enable all kinds of other data-driven decision-making,” said Bonachi.
Organizations have been using predictive AI for some time, says Bonaci. “What makes predictive AI even more powerful is the ability to leverage real-time data to enhance instant experiences and recommendations for customers. For example, Netflix not only recommends streaming content you want to watch based on your viewing behavior, but also predicts which artwork you will be most attracted to and personalizes TV and movie title covers in real time.
Prescriptive AI goes a step further than predictive AI, suggesting the best possible course of action. For example, in a medical scenario, computer vision ML algorithms might be used in predictive AI applications to determine which of thousands of medical images are likely to indicate that a patient has cancer. Prescribing algorithms go further by determining the best treatment to offer a patient from the available data. Alternatively, predictive systems can tell businesses a customer’s “churn” score, or how likely they are to discontinue, while prescriptive systems proactively advise steps to be taken to reduce the likelihood that customers will churn.
Another example of a commonly used application is shipping logistics. Predictive algorithms tell human coordinators how long it will take drivers for each route to complete a delivery. A prescriptive algorithm determines the most efficient route each driver should take to complete a delivery quickly and safely.
The main difference is that predictive AI predicts the future based on past (or current real-time) data, whereas prescriptive AI tells us how to shape the future according to our own requirements.
connect the dots
The last few years have seen incredible progress in the AI tools and applications available to businesses, but it’s clear that we’re still in the early days. Many of the future developments we see could come from combining the generative, predictive, and prescriptive elements of AI.
One of the developments we can expect is advances in the “emotional intelligence” aspect of AI. There will be applications that better assess and react to our own emotional states and provide responses that are more empathetic and attuned to our way of thinking.
A great example is the California-based company Uniphore. The company utilizes his DataStax Astra DB to power his AI-driven sentiment analysis platform, which analyzes tone of voice, facial expressions and spoken words in real time during customer interactions. With Astra DB, Uniphore can efficiently capture and process about 200 data points per frame of a meeting participant’s face, as well as analyze voice tones and natural language processing. Through a multimodal approach, Uniphore’s platform provides businesses with valuable insights to help them understand and respond to customer sentiment to improve customer engagement, build trust and improve customer-facing employee performance.
We are also seeing the emergence of “explainable” AI applications. Recognize that trust is a critical factor in driving adoption of AI tools. They explain how and why that decision was made to eliminate the “black box” problem in AI using generative techniques such as natural language generation.
And closely related to explainable AI is the concept of ethical AI apps. They use generative, predictive, and prescriptive methods to ensure that the decisions they make are (as far as possible) unbiased and unlikely to cause harm or damage to society.
The interplay between prediction, prescribing, and generation gives AI developers powerful options when it comes to understanding our intentions, deciding how best to assist us, and delivering output in a personalized and understandable way. This combination enables some of the most exciting AI applications in business today and promises continued innovation in the future.