New Google Bard leak reveals a ton of new features headed for Google's ChatGPT rival

News
By Alyse Stanley
published

More details on upgraded "Bard Advanced" version tied to Google One subscription

Google Bard
(Image credit: Google)

Google has some big plans for its Google Bard AI assistant, but while the company is staying tight-lipped on an official launch date for now, a new leak may have blown the lid on a few of the features we can expect to see. 

This week noted leaker Dylan Roussel shared on X six AI-powered features the search giant reportedly has in development for its ChatGPT rival based on tidbits hiding in the code on Google Bard's website. He also uncovered a screen hinting at an upgraded version called "Bard Advanced" that could be locked behind a paywall. 

According to screenshots Roussel shared, Bard Advanced is described as a “more capable large language model with advanced math and reasoning skills.” Its upgraded capabilities purportedly lie in the power of Gemini Ultra, Google's top-of-the-line language model, and a serious improvement on the mid-tier Gemini Pro powering the current version of Bard. Though it looks like you'll need a Google One subscription to access these improved capabilities. It's not clear how expensive the company could make Bard Advanced within the tiers of Google One.

The first potential Google Bard feature, dubbed "Motoko," could let users "create bots," though what that entails remains unclear, Roussel explained. But executing the code pulls up a sidebar titled "Your Motokos" along with the option to "create Threads," which also doesn't have much in the way of explanation. 

"I can't say what they mean by bots exactly, if they will be shareable or anything," Roussel writes, "but yes... Motoko is the codename for 'Bots.'"

Next up is "Gallery," a feature that lets users "Explore different topics to see what you can do with Bard," presumably designed for those unfamiliar with AI chatbots. The tipster suggests it could be used to display example prompts based on a range of topics and then search the results to give users a taste of Bard's AI capabilities. A similar screen already exists in the beta version of Bard, where opening its page prompts a welcome screen to appear followed by several suggestions on how to get started. 

There's a "Tasks" feature as well, but its functionality remains a mystery for now. Roussel theorizes it could be used to manage long-running processes, such as generating a photo. He also spotted a new "Sharing Background/Foreground" option that's a bit easier to speculate on: It seems designed for users to more easily share their Bard queries by creating a public link as well as customizing the background and foreground that displays along with it. 

Finally, the last potential Google Bard feature in development is "Power Up," which points towards a way users can "improve" their prompts for a better and more precise outcome. Presumably, it'd work by reviewing your prompt and delivering tips to help you expand your search. 

Of course, it's worth noting that this all remains speculation for now, and it's possible none of these features could see the light of day. As for the upgraded version, Google confirmed Bard Advanced's development back in December along with a slew of new details about its next-generation artificial intelligence model Gemini. The company mentioned then that the "Bard Advanced experience" would arrive early in 2024. No further details have surfaced since. 

More from Tom's Guide

Alyse Stanley
Alyse Stanley
News Editor

Alyse Stanley is a news editor at Tom’s Guide overseeing weekend coverage and writing about the latest in tech, gaming and entertainment. Prior to joining Tom’s Guide, Alyse worked as an editor for the Washington Post’s sunsetted video game section, Launcher. She previously led Gizmodo’s weekend news desk, where she covered breaking tech news — everything from the latest spec rumors and gadget launches to social media policy and cybersecurity threats.  She has also written game reviews and features as a freelance reporter for outlets like Polygon, Unwinnable, and Rock, Paper, Shotgun. She’s a big fan of horror movies, cartoons, and miniature painting.

1 Comment Comment from the forums
  • Fairlytaleree
    17:23 UK
    Thanks for this. A internal version of Bard - presumably in development - as a future option set (while I can't presume to know Bard Advanced functionalities now) may include messaging and file transfer for certain user scopes.
    I have reason to believe that I have expressed to Bard (without going into detail on today) corporate and full common usage Bard progressive works and pricing tier(s) could include full scope of Google's prowess in LLM software hardware.
    As it would be for some better that way if they are new to LLM chat input at all, and the future of us working with such systems.
    For your awareness, I believe I understand that Bard is trained on a LOT of data on the historical internet's content (unsure as to this say) and I believe that in Bard's case real time (almost) input to the internet even is considered in it's responses and style. This would make it great to try!

    If you were not aware there was a bug with Bard which caused it to not show previous chats (I have quite a few now) that was fixed in the last few days. For personal reasons I'm not going to be using Bard directly for a while I write here.
    Reply