From the course: Foundations of AI and Machine Learning for Java Developers

Predictive AI vs. generative AI - Java Tutorial

From the course: Foundations of AI and Machine Learning for Java Developers

Start my 1-month free trial

Predictive AI vs. generative AI

“

- Okay, let's drill down a bit on predictive AI versus generative AI. Predictive AI deals with forecasting using structured data. Now, what exactly is structured data? Structured data is data arranged in predictable defined formats. It might be in a series of labeled files, it might be in a spreadsheet, or a relational database with tables, rows, and columns. So, basically, data that has a set scheme or format. Predictive AI deals with structured data, whether it's for recommendation systems or classification, doing business analysis, or fraud detection or predictive maintenance. An example of predictive AI is weather prediction. You're forecasting information based upon weather information that is structured data from the past. Predictive AI has been used for many years in financial markets, certainly investment banking where I have some experience. For many years, it has used historical stock or bond prices to predict future movement. Predictive AI can be used for spam control. There are companies that use predictive AI to detect patterns in your email and determine whether those patterns indicate if an email is spam or not. There are a lot of use cases for predictive AI and forecasting the future based on structured data. Generative AI is where you're creating things based upon unstructured data. This type of data does not have a predefined format or organization. There's no schema involved in the data. The data is very freeform. You're dealing with data that's not in the table or a spreadsheet or in a label file. Generative AI is really, really good at handling this type of data. For example, emails, social media posts, images where there's no specific structure to the data. Generative AI is a great tool to use to analyze this type of data and look for patterns. You might want to use generative AI to generate suggestions or ideas based upon a large set of documents, or you might want a summary of a very large financial report listening to the important points raised in that report. Or you might want a document just translated into another language or reword it for a younger reader. And here are some examples of where you might use generative AI. You might have some text that describes your company or your organization, and you want some business ideas related to that text. Or you're a teacher and you want to design a curriculum on a certain topic and you want a basic curriculum designed to get started. Or you might want to have generative AI look at your emails and give you a summary of those emails. And maybe perhaps you might want to rank those emails in terms of importance. You might want to say, "Give me a summary and rank them from 1 to 10 so I have an idea which emails I should respond to sooner rather than later." There are many places for generative AI and many places for predictive AI. Typically, they're used for different use cases. They're not mutually exclusive. You can also combine them. For example, you can use predictive AI to get customer preferences from your customer. Let's say, for an e-commerce application. And then use generative AI to create a specific plan for those customers. Very personalized plan to address those customers. Another example is that you're detecting fraudulent financial transactions with predictive AI. Then you say, "I want to use generative AI to come up with ideas to counter that fraud." Now, there's a slight overlap between the two, but primarily one is for forecasting and the other is for generating new content. And again, you may want to use these two different techniques in a powerful combination.

Contents