Machine learning watching generative artificial intelligence (AI) take off feels a little bit like an American Girl doll envying Barbie for getting a movie. What is she, chopped liver? Like the American Girl doll, we can’t forget about machine learning, because it’s the giant that generative AI is standing on. How do they relate? Well, machine learning is how generative AI learns. . . but generative AI takes machine learning a step further by leveraging those learnings to produce something new. Here’s an explanation of the relationship between generative AI vs. machine learning from the Everest Group:
“...Generative AI is an emerging technology that has the potential to revolutionize the way businesses operate. It uses ML and deep learning algorithms to generate new and unique outputs, such as text, images, or even music, based on a set of inputs and a trained model.”
When it comes to generative AI vs. machine learning, think of AI as an umbrella term for all types of AI, including generative AI. Machine learning is how any artificial intelligence learns. Similarly to how there are many types of AI, there are also plenty of machine learning models, such as transformer models, diffusion models, or generative adversarial networks (GANs).
In this article, we’ll look at a use case—processing email correspondence—in two parts to see where machine learning comes in to support generative AI. This use case, which applies to pretty much any organization, can help illustrate how AI can support and enhance business operations.
8 experts weigh in on the future of AI. Download their insights in the 2024 AI Outlook.
An example of generative AI vs. machine learning at work.
Organizations receive a constant influx of correspondence—from customers, prospects, partners, vendors, etc.—and they always need to process it. Although it’s painstaking and never-ending, it’s a highly important aspect of the operation. And it’s an excellent candidate to hand off to AI.
To start processing emails with AI, the organization first needs to classify them. Next, the organization may want to take action by generating some notes and/or sending a response that acknowledges receipt of the email or provides information to the sender. Both of these steps are excellent opportunities to use machine learning and generative AI.
To make this example more concrete, we’ll look at how you can operationalize AI by building an AI Skill in the Appian Platform.
Part one: Classifying emails with machine learning.
To classify the documents, the developer needs to create an AI model. This is where machine learning comes in. Here’s how it works. First, using the Appian AI Skill design object, the application developer selects “email classification.” Next, the developer uploads a suitable set of emails to train the model, and waits while it does the necessary training. This is machine learning in action! The model is taking the emails and learning from them to understand the different ways it could classify things. For example, it’s learning which emails should be classified as a customer support email, which emails are sales related, which are press inquiries, etc. Once it has learned, the developer designing the skill can test the model’s categorization abilities, determine whether they are accurate enough to be used, and if so, add this new skill into their application to begin processing content.
Part two: Writing email copy with generative AI.
After classifying emails, employees need to respond to them. So, the application developer connects the application to the Appian OpenAI plug-in to generate content for the emails. (Here comes generative AI!) This plug-in feeds the sender’s email copy into the generative AI tool, gives some tips for how to respond as a prompt, and allows the plug-in to generate that initial response. The employee can edit the generated email as needed.
So that’s how one application might use both machine learning and generative AI in different ways. In part one, machine learning helped the model understand how to classify emails, and in part two, generative AI aided the employee in responding to the sender.
Want to learn more about the future of artificial intelligence and hyperautomation? Get the Gartner® Emerging Tech Radar Hyperautomation Report.
Generative AI vs. machine learning: partners for transformation.
As you can see, generative AI wouldn’t be much without the help of machine learning. But machine learning has come a long way in order to now enable generative AI. They’re really not opposed to each other at all but rather partners that your organization can use to become more efficient. (And maybe the American Girl dolls can hold out hope that they’ll get a movie someday, too.) In our email processing example, anyone who has to respond to emails in your organization would benefit from the assistance of machine learning and generative AI to make this task easier and faster.
There are many use cases for applying generative AI to business practices. Learn more about the differences of generative AI and large language models to continue exploration around AI.
Also, we didn’t get into all the ways you can optimize content processing with AI, but there’s definitely more there. You can watch an expert go deeper into this use case below.
