AI EDUCATION: What Is Predictive AI?

Most of the words we’ve written about artificial intelligence so far in AI Education have been about generative AI, which is a relatively new, emerging technology, really only in the zeitgeist for about three years. But AI has already been used across industries, even to some extent in tech-averse and highly regulated fields like financial services, for most of the 21st century. However, most of the artificial intelligence we’ve used to this point has been predictive AI, built as an extension of predictive analytics, and is distinct from the more newfangled generative AI. 

Predictive AI is technology consumers encounter every day in online marketplaces with variable pricing, or on applications that make recommendations for products or services. Financial services professionals are usually using predictive AI when their software recommends next-best actions or gradually “learns” to prioritize and bring forward the most used tasks and applications, or to move towards a user’s next action within a workflow. However, predictive AI can be  used for much more sophisticated and consequential tasks than a recommendation engine. 

While generative AI is more of a blue ocean where new applications and other possibilities are still being discovered, there’s still a lot of progress to be made on predictive AI as well. More importantly, predictive AI and generative AI complement each other and synergize in powerful ways with implications for the financial services workplace. 

What Is Predictive AI? 

Predictive AI combines statistical analysis with machine learning, giving software the ability to identify patterns, forecast events and anticipate actions and behaviors. It can fit any place where data-based decisions are being made, increasing efficiency and accuracy by analyzing large volumes of data in order to reach an optimal conclusion. 

Think about all of the market analysts earning six-figure incomes by poring over historical data—whether it be the relationships between earnings and other financial data and stock prices, or understanding different chart patterns, or the impact of policymaking on the fixed income market, a lot of people have been employed for a long time to focus on what was very granular minutiae on individual companies in hopes of divining what was going to happen in the future, all in order to make a little bit more money. Predictive AI promises to do the same kind of work much faster and, hopefully, more accurately at a tiny fraction of the expense. 

Armed with a wealth of accurate data, predictive AI can be used to understand complex behavior, aiding in business decision-making. It can help business predict issues with customers, supply chains and equipment. Artificial intelligence, like even the best human analyst, is not infallible, and it has no crystal ball. When predictive AI is asked to say something about the future, it always speaks in potentials or probabilities, never certainties. 

How Does It Work? 

Without getting too much into the weeds, predictive AI uses machine learning to pore through large volumes of information using various machine-learning and AI algorithms. Neural networks may be employed to help identify patterns within the data. Different techniques are used to classify and organize information and understand the relationships between different pieces of information. For example, linear regression algorithms help find correlations between different pieces of data. 

By “training” on data, algorithms learn historical correlations between different variables that precede certain outcomes, and then are able identify those correlations as they occur in real time, which enables them to make predictions. As more data is collected over time, and more real-time predictions are made, a predictive model can “learn” from its own activity and refine itself, becoming more accurate and successful. 

Applications for Predictive AI 

Inventory management—predictive AI can help inform a business of potential changes in consumer demand. 

Supply chain management—predictive AI can help businesses understand when traffic, weather, conflict or other conditions will disrupt supply chains. 

Personalization—predictive AI can anticipate client needs and tailor services to their preferences. 

Healthcare—predictive AI can forecast health conditions based on a person’s medical history. 

Insurance—predictive AI is used to forecast potential losses and to make the claims process more efficient. 

Finance—predictive AI can be used in fraud detection and to predict market movements 

Predictive AI Versus Generative AI 

Predictive AI uses machine learning and historical data to extrapolate the potential future. Generative AI, on the other hand, uses large language models to create content from prompts written in natural language. One attempts to look ahead by reading historical data, the other attempts to create meaningful new data from reading historical data. Predictive AI is like the analysts at a financial firm, generative AI is more like the media or public relations director. 

Generative AI incorporates some of the same underlying technology as predictive AI—in fact,  the portion of generative AI that “reads” a user’s query and then translates a response into text, video, images or other media, is using predictive AI to best understand the request and to predict the best possible response to fulfil that request. 

A crucial difference between predictive and generative AI models also lies in explainability, the ability for AI to demonstrate and make clear how it arrived at its results. For predictive AI to be deployed, it must deliver explainable results so that users can go back and understand why its output was accurate or inaccurate. Generative AI, on the other hand, usually does not have the same explainability requirements, except perhaps in heavily regulated spaces like healthcare and financial services.