Google teases a Gemini feature that will finally make it a ‘true AI assistant’

Google teases a Gemini feature that will finally make it a ‘true AI assistant’


Summary

  • Google’s upcoming “Research with Gemini” feature aims to become more advanced, helping users with reasoning, planning, and memory like “a true AI assistant.”
  • Gemini will save time by researching information across the web and creating tailored reports for users’ specific questions.
  • The promising features of Research with Gemini include being able to think ahead and complete tasks on behalf of users, evolving into a helpful researcher.




AI had a huge year in 2023, but growth seems to have stagnated in 2024 as companies find themselves searching for genuinely new and useful ways to implement it. Google’s Gemini AI had a big day on the stage in Mountain View earlier today, as the company used it as a framing device for all of its Pixel hardware announcements. Most of these features are debuting today on the Pixel 9 , but at the very end of the presentation, VP Rick Osterloh came back out to tease Gemini’s next chapter.

Related

Google Pixel 9: Everything you need to know about the new flagship series

Google’s biggest mobile shakeup since Tensor is almost here


Per Osterloh, “We’re evolving Gemini to be even more agentive, to tackle complex problems with advanced reasoning, planning, and memory.” Planning and memory are the two words to focus on there: Gemini with the ability to plan could potentially anticipate what you’ll need it to do next, preparing an answer before you ask. And a proper memory feature could give Gemini the medium- to long-term context needed to help with bigger projects.

Osterloh explained that “You’ll be able to think multiple steps ahead and Gemini will get things done on your behalf, under your supervision — that’s the promise of a true AI assistant.” What made our ears perk up was that “under your supervision” line — if Google is already trying to get ahead of privacy or legal concerns months before debuting this feature, it may actually be groundbreaking enough that people won’t fully grasp how it’s working at first.

Gemini will get things done on your behalf, under your supervision — that’s the promise of a true AI assistant.


The entire Made By Google keynote is embedded below, but the video should start playing right where Osterloh begins talking about Research with Gemini:

At the very least, it sounds like this feature has the potential to shake up how school projects are researched, making Gemini the type of homework helper that could give teachers fits:

Soon, Gemini will be able to assist you as your researcher, saving you tons of time by using information from across the web to create a research report that’s tailored to your exact questions. What used to take you hours now takes minutes.

One example Osterloh gave was getting help with researching grad school scholarship programs in certain fields. Another asked Gemini for help with researching the process of opening a sidewalk café in Seattle.


In the teaser, Gemini then created a multistep research plan with one-tap shortcuts for reading up on the key points. Osterloh said that Gemini would then curate information from all over the web, even going so far as to navigate drop-down menus and sub-pages on your behalf to surface finer details.

Render animation demonstrating how Research with Gemini might work.

Source: Google

Google showed a brief demo of what this functionality might look like once it’s ready for the public. Osterloh said that once the initial research prompts have been processed, Gemini will put together its research findings in a Google Doc. In its example with the Seattle sidewalk café, Google envisioned the tool looking into local permitting and regulations, then summarized its findings in the Doc, complete with expandable links to the sources.


But while Osterloh emphasized that this tool “isn’t a far off feature at all” and would be launching to Gemini Advanced users “in the coming months,” Google flashed a series of footnotes at the end of the livestream, and the one for this feature specified that the UI shown in the demo is not finalized and remains subject to change.



Source link

Comments

No comments yet. Why don’t you start the discussion?

Leave a Reply

Your email address will not be published. Required fields are marked *