Bard working in Google Messages

 Google Messages may get some Bard-powered features

Google is integrating its AI chatbot, Bard, into Google Messages, allowing users to start RCS chats with Bard directly from the "New conversation" screen.  Bard will presumably serve as an all-in-one helper in Messages, offering features such as drafting messages, image identification, language translation and book recommendations based on location and past chats.  Bard chats are not end-to-end encrypted, and trained reviewers may review conversations to improve the chatbot.

Over the past year, Google has been diving deep into AI, integrating the Bard chatbot into various apps and services like Google Workspace, which includes Docs, Gmail, Drive, Sheets and more. Now, it looks like Google Bard is making its way to Google Messages, bringing its helpful AI features to more users. Google is integrating its AI-powered chatbot, Bard, into its default messaging app on Android, Google Messages. Bard in Google Messages will help users write messages, translate languages, identify images and explore interests. The Bard-powered features in Google Messages are similar to those available in Google Allo. Google Messages users can initiate Bard from the New Conversation screen, where it appears as a standalone chat. Bard can draft messages, identify images, suggest books and recipes and more. In this integration, Assistant existed both as a standalone conversation as well as a tool that users could invoke inside other chats. The company shut down its Allo app in March 2019.

Google caught a glimpse of how the AI chatbot will work in the newest beta version of the Messages app several months after initial hints of its existence in the app were spotted. It looks like you'll be able to start RCS chats with Bard in Messages straight from the "New conversation" screen. It's not your usual chat; instead, it's a standalone one with the Bard logo and name up top, as if you're chatting with a virtual "penpal," which also happens to be the codename for the feature. Based on a few code snippets, it seems Bard is gearing up to be your all-in-one helper in Google Messages for things like drafting messages, image identification, language translation, book recommendations etc. Apparently, Bard learns from your location and past chats to give you spot-on answers.

When Bard whips up a reply, you can give it a thumbs up or down with a long press. You can also copy, forward, and star the response. Unlike your typical RCS chats, Bard chats aren't protected with end-to-end encryption. Google says these conversations are used to improve its services, especially the brainpower behind Bard. This means trained reviewers might take a peek into your conversations. So, if you don't want a reviewer to see your chats, steer clear of sending any confidential or sensitive messages to Bard. The data that gets reviewed is separated from your account and remains in place for a maximum of three years, according to the description.

As per the report, the Bard-powered features in Google Messages are now available with the latest beta version of the app. The report explains that Bard in the Messages app can be initiated from the “New conversation” screen which appears as a standalone chat with the Bard logo and name appearing at the top in place of an actual person. This integration is codenamed “penpal.” Google Messages users can also access Bard to draft messages, identify images, suggest books by genre, offer recipes given a list of available ingredients and more. The report noted that the company has also shared a few example prompts which were:-

“Write a text message I can send to my boss calling in sick today. It’s a big team presentation day, so I’m sad to let everyone down. 3 sentences or fewer.”

“Create a vegetarian meal with the following ingredients I have in my fridge: cauliflower, cucumber, and yogurt”

“Come up with a complex word riddle that has the answer: sunset. You are a skilled riddle maker, known for the cleverness of your wordplay and the complexity of your puzzles, skilled at thinking up clever and challenging riddles. Be creative. Come up with a riddle that is difficult to guess. Include hints to help the user guess.”

As per Google, Bard chats and their data are saved for 18 months. Even if you decide to disable Bard activity, the data's still hanging around for 72 hours. You can manually manage or hit the delete button on this data. This feature isn't live in the app just yet. It's entirely possible that after some trial and error, Google might offer the idea to all users. On the other hand, this could be the next big thing for tools like Smart Reply, giving you those nifty response prompts based on your chat context. Bard's not just a machine-learning model for snappy replies; it's more of an AI chatbot. Baking it into a messaging app opens the door for it to take the reins on your chats. Now, before Google lets Bard run wild with the replies, the company would probably need to train it a bit more. As a result, users might have to dish out a bit of personal information and text history, so the AI can get the hang of responding just like them.

The report also notes that whenever Bard generates a response, it’ll be marked with a sparkle next to the time/date. Users can long-press to give it a thumbs up/down, as well as copy, forward and star. However, on the privacy front, Google said that the “chats with Bard are not end-to-end encrypted.” The company added these chats will be “used to improve Google services, including the machine-learning models that power Bard. As part of that improvement, trained reviewers need to process your conversations, so when using Bard, don’t enter anything you wouldn’t want a reviewer to see or Google to use. Reviewed data is disconnected from your account and retained for up to 3 years.”