Google’s Smart Reply, a feature that suggests smart responses to a received email, is coming to Gmail on Android and iOS, Google announced on Wednesday. The feature was already available on the Inbox by Gmail application and Allo, Google’s smart chat app.
When a user receives an email, Smart Reply presents them with three responses to choose from, based on the email they received. For example, if the email suggests meeting at a date and time, one of the responses might mention if that date and time is good or bad for the user.
According to a press release announcing the feature, users can choose to send one of the provided replies immediately, or they can edit and customize the message before sending.
Because Smart Reply relies on machine learning, the more users make use of the feature, the better its responses will be. For example, if you tend to send short replies, like a simple “Thanks,” the service will eventually begin suggesting more replies that follow that style, the release said.
Smart Reply for Gmail will initially roll out globally in English, but will be available in Spanish a few weeks after that, according to the release.
Google’s work with artificial intelligence (AI) and machine learning continue to seep into its work in business technology. In fact, the company is even using AI to help other companies fill jobs, and to help professionals detect objects in videos.
As reported by TechRepublic’s Nick Heath, Google seems to be pursuing narrow, focused AI that is good at specific tasks, as opposed to AI that is more like general human intelligence. And as Google’s DeepMind continues to beat champion Go players and improve its intelligence, one can only wonder what Google’s next AI-powered tool or service will be.
The 3 big takeaways for TechRepublic readers
- Google is rolling out its Smart Reply feature to Gmail for Android and iOS.
- The feature offers set responses to a given email, allowing users to select one and send it, or edit it and then send it.
- Using machine learning, the tool will begin to predict responses that are more in line with what the user would actually say.