I have tried a lot of “AI for WordPress” plugins over the last two years, and most of them were bloated wrappers around an API plus a messy sidebar in wp-admin. AI Experiments 0.3.0 reads differently. It comes out of the canonical AI initiative, so the features are built into the core editing workflows instead of bolted on beside them.
The 0.3.0 release is a refactor rather than a patch, and it ships actual editor UIs for content summarization and image generation. If you read my earlier notes on the WordPress AI Client architecture, this is where those “Abilities” stop being a backend concept and start showing up where writers work.
What’s new in AI Experiments 0.3.0?
Most of the work in this release went into how content is handled. The direction has shifted away from “chatting with your post” and toward blocks and sidebar panels that each do one job. Three experiments carry that change.
1. Content summarization and the AI Summary block
Rather than generating a summary and leaving it in a transient, AI Experiments 0.3.0 lets authors generate it in the editor and place it in a dedicated block. Because the text then lives in the post content, themes and feeds can read it without another API call on every page load, which helps accessibility and search engines alike.
2. Featured image generation with metadata
Generating an image is easy. Handling the media properly in WordPress is the hard part. Image generation now sits in the post editor sidebar, and it does more than drop a JPG into the library: it derives the prompt, generates alt text when that experiment is switched on, and writes proper AI attribution metadata.
3. Alt text generation for accessibility
Accessibility tends to be the first thing that breaks as a site grows. The Alt Text Generation experiment has been tightened up, and authors can now ask for a description from the Media Library or from the Image block itself. Nothing fires on its own; a person has to request it, which keeps invented descriptions out of your alt attributes.
Under the hood: developer improvements
If you read the source, 0.3.0 is mostly boring housekeeping, which is what you want from a plugin like this. The Abilities Explorer got strict typing, asset handling improved so a missing file no longer throws admin warnings, and the global “Enable Experiments” logic got simpler.
If you want to work with these abilities in code, the tooling is moving toward @wordpress/build and DataForms. This is roughly how I check that an ability exists before hooking into a workflow:
// Example: Checking for a registered AI ability in a custom block
function bbioon_check_ai_capability() {
if ( ! class_exists( 'WP_AI_Abilities' ) ) {
return;
}
$abilities = WP_AI_Abilities::get_instance();
if ( $abilities->is_registered( 'summarize-content' ) ) {
// Safe to inject your custom summarization logic or UI
}
}
add_action( 'init', 'bbioon_check_ai_capability' );As I said in my critique of the AI Client, the point of all this is a standardized interface. 0.3.0 gets closer, and the TypeScript and DataViews refactor planned for 0.4.0 should get closer still.
If wiring up AI Experiments 0.3.0 is eating your dev hours, hand it to me. I have been wrestling with WordPress since the 4.x days.
Where I would use it
Do not drop this on a production site and walk away. The name says “Experiment” and it means it. For agencies and developers building editorial workflows, though, AI Experiments 0.3.0 is the first WordPress AI tooling that has felt like part of WordPress rather than a bolt-on. The release is on the official plugin repository, and the full changelog is on GitHub.
