Google is currently testing a new feature called "Sparks" within its Illuminate platform, which focuses on generative AI content creation.

As Google I/O approaches, Sparks is expected to introduce significant updates to AI media, reflecting the company's commitment to innovation.

Sparks generates short, vertical explainer videos, lasting one to three minutes, based on a single prompt, and is designed in a TikTok-like style.

The AI behind Sparks is capable of producing voiceovers, scripts, and synchronized visuals, showcasing a sophisticated multimodal model.

This feature could redefine how Google approaches content consumption and creation, emphasizing the simplification of complex information through engaging visuals.

Illuminate has evolved from providing audio discussions of research papers to a broader media generator, now including editable AI summaries and image generation tools.

Currently, there is no public release date for Sparks, as it remains an internal testing project.

The rapid development of Sparks raises questions about the number of generative AI products Google is developing behind the scenes.

Sparks may utilize Google's Gemini multimodal models or the Veo 3 video generation system, both of which are still under development.

The high quality of Sparks videos suggests a connection to advanced AI models like Veo 3 or a multimodal variant of Gemini, indicating powerful underlying technology.

Overall, the developments within Illuminate signal Google's direction towards integrated multi-modal content generation from single prompts.

Sparks overlaps with Google's NotebookLM initiative, which is also set to include AI-generated video explanations by virtual presenters.