Exploring the Future of Web AI with Google

What if browsers shipped with built-in AI models? Learn how this shift to on-device AI will change web development and eliminate massive model downloads.

#1about 2 minutes

Introducing Google's fund for web AI innovation

The Google AI fund supports developers building on-device, cloud, or hybrid AI solutions to encourage ecosystem growth.

#2about 2 minutes

Optimizing on-device AI with right-sized models

Avoid performance issues by selecting the smallest model that delivers acceptable quality for the specific use case and device.

#3about 4 minutes

Using built-in browser models to reduce overhead

Browsers can provide built-in models for common tasks, which reduces redundant downloads and offers developers simple, high-level APIs.

#4about 4 minutes

Standardizing web AI APIs across different browsers

New AI capabilities are being standardized through proposals like navigator.ai, with developers using polyfills to provide cloud-based fallbacks in unsupported browsers.

#5about 4 minutes

Navigating the web AI API release process

New web AI APIs become available through a phased rollout, starting with developer flags and origin trials before being shipped in stable browser releases.

#6about 4 minutes

Competing with built-in browser AI user interfaces

Developers face a new challenge as browsers increasingly add their own AI-powered UI, which can compete with or interfere with custom application features.

#7about 8 minutes

How developers can prevent browser AI interference

Developers may need new HTML attributes or APIs to prevent default browser AI behaviors from conflicting with their own custom implementations, similar to disabling autocomplete.

#8about 5 minutes

Connecting web apps to local OS-level AI models

To avoid redundant downloads, web applications can potentially access local OS-level models through methods like local web servers or future browser-OS integrations.