The Role of AI & Machine Learning in the Future of Content Creation

Let's get one thing straight up-front: This blog was not written by an AI. (But full disclosure: The teaser was).

And let's be honest. Since early 2023, AI has become a household term.

It's not that AI just suddenly entered onto the scene. In fact, it's been chugging along and improving capabilities for years. But all of a sudden - within just a few months - literally hundreds of millions of people have suddenly become aware of the capabilities and potential of AI, and millions of people have started to actually use it in meaningful ways.

In response to this sudden awareness, many vendors have quickly rolled out new AI features for their products. While many of these are useful, some of them are fairly cosmetic (giving a sort of "look, we're not behind!" vibe).

dotCMS is also very aware of both how much attention AI has gotten and of the capabilities of the tools. However, instead of just jumping in with a "me too" feature or two, we've spent some time digging deep into both the technologies and the dotCMS platform to effectively plan how to enable our customers to make the best use of these technologies with their own content in dotCMS.

Although we've already defined a few technical terms, we're not going to try to explain in depth what the technologies are or how they work. What we will do, however, is share with you our view of these technologies. This includes, first, how we view the capabilities of AI in the context of the content management space. Second, what our strategy is for incorporating those tools into dotCMS to enable you, the users, to build amazing (and award-winning) stuff with dotCMS.

AI & Machine Learning in Content Management

What AI allows us to do is more important than how it works

Uses of ML technologies are rapidly evolving, and people are finding new ways to use ML technologies literally every day. And there are thousands of ML-driven applications available right now, which use ML technology to provide services for everything- from marketing to business analysis to software and hardware development.

People categorize the types of things you can do with ML technologies in different ways. While common classifications often make sense from a technical perspective, they don't always make sense from a practical perspective. For example, generative AI can be used to both create content from whole cloth and re-write existing content, but from a user perspective, these are quite different activities.

So, we've outlined our own set of categories for how AI capabilities can help specifically within the content management space. These categories aren't defined based on what technologies are used; instead, they identify several key types of work these technologies can enhance.

In other words, these are areas where ML technologies can help developers and content editors deliver better results faster. These are the areas we're focusing on as we build ML technology into dotCMS.

Search: More natural and/or more accurate content search and retrieval

Traditional content search based on specific search terms often requires some expertise to get the desired results and can also be quite limited in the results returned. In contrast, ML systems can allow users to ask questions in natural language, can take more context into account (without the user having to enter it), and can provide results which aren't an exact match, but which use similar words and phrases.

Generative: Generate new content for your sites and applications

This can include generation of text, images, video, code, or data and usually requires some kind of prompt to be sent to the ML system. The majority of the recent attention on AI technologies is centered around generative text tools such as ChatGPT and Google Bard, as well as generative image tools such as DALL-E and Midjourney.

Note, however, that even in this fairly straight-forward use of ML technology, there's plenty of room for innovation in how the prompts are managed. For example, the prompt sent to the ML system can be user-supplied, have context added to it behind-the-scenes, or automatically generated in some fashion.

Transformational: Modify existing content

This is actually a form of generative AI. However, instead of having the ML system create brand new content based on some prompt (which may require you to learn the newly-minted skill of "prompt engineering"), you ask it to modify some existing content based on specific criteria you give it.

For example, one common use of this is to send an ML system an existing block of text and simply instruct it to "Make Shorter" or "Make Longer." You can also get more sophisticated, such as asking an ML system to rewrite text or re-generate images in a specific style, to ensure all your content conforms to some consistent style or branding.

Analytical: Analyze and summarize content

ML systems can analyze content, which can be used to describe or summarize it. For example, ML technologies may be able to summarize the content of a page for SEO purposes, provide summaries of documentation for your users, or automatically generate image descriptions for alt-text tags.

Predictive: Suggest content values or user actions

This is a form of Analytical AI, specifically tailored to generate suggestions based on the user's current context. This context can include such things as the content the user accesses, the actions they take, and user or system history.

For example, dotCMS has a popular tag suggestions feature that suggests possible tags for content when a user begins typing in the Tags field. This feature is currently based on simple content search but could be enhanced to use machine learning to analyze the content and suggest tags based on the text in the content - even before the user begins typing.

Integrative: Using ML in ways that are adapted to specific features

This simply means using ML technologies in a way that's specifically tailored to improve or simplify the use of existing product features. This could be done using any of the above methods, or some combination, but requires a deep understanding of both the product and how it's used. In many cases, integrative uses of ML technologies will be hidden from the user. It will improve their experience, without them having to explicitly choose to use "AI," or even know it's happening.

For example, dotCMS has several personalization features, including visitor personas. ML technologies could potentially be used to automatically assign a persona to a site visitor, or allow content editors to automatically generate different versions of the same content for different personas. It could even potentially generate a different version of the same content for each individual user, based on their past history.

Read more on Artificial Intelligence & Content Management Systems: A Forward Thinking Use Case.

Our AI Roadmap

Tools, not toys

With all this in mind, let's talk about how dotCMS is going to empower you to use these powerful new technologies with your own content.

To be frank, we don't want to just "bolt something on" that doesn't integrate well with the whole product. Our goal is not to create features for appearances. We're not interested in creating toys that look flashy and demo well but don't help you get your work done.

Instead, our goal is to create tools that enable you to use these new technologies on your own content, and in your own sites and applications. We want you to be able to use these technologies to help everyone who interacts with your sites and applications- from your developers to your content creators to your end users.

Phase 1: Foundation and Content Generation

So, the first phase of our ML implementation (available in Q4 2023) will do two things. First, it will build a foundation that allows you to access these capabilities not just in one or two places, but from almost anywhere in your site or application. Second, it will provide some specific tools to make it easy for your content creators to generate content with these technologies within some of our existing tools.

The foundation will include both an API and a Velocity viewtool. These will allow you to easily access AI technologies from a wide range of dotCMS features, including Themes and Templates, Containers and Widgets, and in your own code and applications via Custom Content Fields, Custom Workflows, and your own scripted APIs.

This foundation will initially be built to use a specific AI vendor (OpenAI). However, it provides standard methods and arguments that don't depend on any specific AI vendor. This means that if you want to change the vendor later, you don't have to change any of your code; all you have to do is replace a plugin.

The specific tools included in phase 1 will be custom blocks for the Block Editor, including an AI Text block and an AI Image block. These will allow you to generate and regenerate content from a prompt, and the AI Image block will allow you to automatically generate an image based on the other text within your content, all without even entering a prompt.

Later Phases: Additional Vendors, Tools, and Integration

In following phases, we'll allow integration with other vendors of your choice, add additional generative and transformational AI features, and integrate native access to ML technologies into many other dotCMS features. When we say integrate, we mean it. We won't just add "new" features for you to learn, but we'll extend the power of the existing dotCMS feature.

We'll work to make it as seamless as possible. Much of the time, you shouldn't even realize you're using "AI" at all; you'll just end up getting more and better work done even faster than you were before.

Coming Soon

Exciting changes are in-flight!

You can look forward to some exciting changes coming (very) soon within dotCMS. The kind of changes that will allow you to gain quick benefit from these new technologies, and also go deep to integrate them into your own sites and apps.

We hope you're as excited about using these powerful tools within dotCMS as we are to deliver them to you!