Google is using machine learning models to speed up Gmail search


Posts: 223   +29
Cutting corners: Introduced back in 2004, Gmail gave users a more robust alternative to legacy web-based email services such as Yahoo and Hotmail. Today, Gmail is one of the most relied-upon email services in the world, accounting for more than 27% of all email opens and more than 330 billion total emails on any given day. That's a lot of email to sift through, and it's exactly the kind of problem Google wants to help you overcome through the power of machine learning.

Millions of users around the world rely on Gmail as their primary email service. And it's no mystery that the sheer number of emails sent and received on a daily basis can make keeping a clean, orderly inbox nothing more than a dream. But help is on the way, according to Google's Workspace Updates team, detailing Gmail's efforts to incorporate machine learning models to enhance its mail search experience.

The blog post indicates that Google is introducing a feature that will harness its latest machine learning technologies to deliver more accurate Gmail search results. This technology aims to provide a more targeted and precise set of results.

Once in effect, Gmail users initiating searches will be automatically assisted by Google's newly integrated machine learning models. These models will consider the user's search terms and recent emails to yield more relevant results that closely align with the user's search query. The blog post states that AI-assisted results will be presented at the top of the results in a dedicated section, followed by all other typically returned results, arranged by recency.

The new feature is being rolled out to all Gmail users with no additional actions or configurations required. Mobile users will simply use the Gmail search bar as normal, entering any keywords required to retrieve relevant and desired search results.

The new Gmail search functionality began rolling out via Google's Rapid Release and Scheduled Release cycles on June 2, 2023. Rapid Release makes new features available to consumers as soon as they are rolled out by Google, resulting in no evaluation period and immediate availability. Scheduled Release, Google's default release setting, delays the release of new features so that admins have time to train staff or prepare their organization for the pending change. The Scheduled Release cycle typically begins 1-2 weeks after the Rapid Release deployment cycle.

Permalink to story.



Posts: 1,391   +2,633

I think this story messed up. Isn't it supposed to be "AI"? AI sells! Not machine learning. No one these days understands what machine learning is, they only know AI.


It's AI!

A f'ing I!

Eh, who am I kidding, there is no AI. Machine learning is actually a welcomed term here instead of the ride the money train use of AI.