Google Chrome’s AI team has proposed new APIs, dubbed Writing Assistance, for directly accessing language models built into web browsers and operating systems. It could lead to more sites and web apps adding AI writing integrations.
Large language models (LLMs) are all the rage thanks to OpenAI’s ChatGPT, which sent the entire tech industry into an AI frenzy and scrambling for similar features. Creating AI web apps that implement LLMs is easier said than done, though. Software developers can move all the heavy lifting to cloud APIs, which cost money, or try to set up local AI models with existing technologies such as WebAssembly and WebGPU. Thomas Steiner, Google’s Developer Relations Engineer, announced on August 19 an effort to provide a better solution, calling it the first step “toward potential standardization.”
The Writing Assistance API proposal, available on GitHub, comprises APIs dedicated to summarizing input text (Summarizer API), writing new content based on a prompt (Writer API), and rephrasing input text (Rewriter API) in the requested ways. Google is providing the APIs as separate proposals so anyone interested can evaluate them independently of one another.
The APIs have not yet been approved to ship in Chrome. If they end up being standardized and adopted by developers, however, web apps will instruct your favorite web browser to download specific underlying AI models for writing assistance on the fly “so as not to lock implementers into a single strategy.” This should make AI-powered web apps more flexible in leveraging language models to provide text summarization, rephrasing, and rewriting capabilities. The company broke down the most common use cases for each API.
Summarizer API
- Creating meeting transcripts for anyone joining the meeting late.
- Summarizing support conversations for input into a database.
- Distilling product reviews into a sentence or paragraph.
- Summarizing long posts to help you decide if it’s worth your time.
- Generating compelling article titles as a form of summarization.
- Summarizing questions on Q&A sites for experts to answer.
Writer API
- Explaining structured data like poll results over time, bug counts by product, etc.
- Expanding lists of pros and cons into full reviews.
- Generating a biography from an author’s CV, list of previous jobs, etc.
- Drafting articles from bullet points or your stream-of-thought.
- Composing social media product posts.
Rewriter API
- Removing redundant or unnecessary information from writing.
- Make your writing more or less formal for specific audiences.
- Rephrase toxic language to be more constructive.
- Rewriting articles using simpler words and concepts (“explain like I’m 5”).
Google says the proposal could allow web apps could rely more on the hardware capabilities of your device, and less on servers, making them potentially faster and nimbler. The functionality could also work without an internet connection. We’ve grown so accustomed to offline support in web apps like Gmail and Google Docs that we take them for granted, but most AI web apps today are crippled without internet access.
Apple has emphasized the importance of encryption and on-device processing with its own Apple Intelligence features coming this fall. In contrast, Google’s proposal promises to combine the best of both approaches. The company also mentions the Writing Assistance APIs would lower API costs for web developers and permit hybrid approaches where, for example, people who use a web app for free could only access a basic AI model (whereas paid users would be given access to a more powerful API-based model).
Chrome’s AI team will be soliciting feedback in the coming weeks, and they’ll ultimately move the proposal over to the Web Incubator Community Group (WICG) for further discussion. The WICG is a forum for discussion, not a standards body, so the project will still need to clear a few hurdles before getting submitted to the World Wide Web Consortium (W3C) for final approval. The WW3C will either turn the proposal into the official specification or reject it. If accepted, it’s up to browser vendors like Google, Apple, Microsoft, Mozilla, and others to implement the whole specification or select parts of it.
