A new iteration of Google's leading AI model illustrates the company's vision of AI revolutionizing personal computing, web search, and potentially the way individuals engage with the physical world. In December 2023, Google introduced Gemini in an attempt to close the gap with OpenAI, the company responsible for the highly successful ChatGPT. Although Google had made significant investments in AI and achieved important research advancements, OpenAI was praised as the new frontrunner in AI, with its chatbot considered a potentially superior method for web searching. With the Gemini models, Google now provides a chatbot as proficient as ChatGPT and has incorporated generative AI into its search and other products.
When Hassabis initially introduced Gemini in December 2023, he informed WIRED that its training to comprehend audio and video would ultimately be revolutionary.
Today, Google provided a preview of how this could unfold with a new iteration of an experimental project named Astra. This project enables Gemini 2 to interpret its surroundings, as seen through a smartphone camera or another device, and engage in natural, humanlike conversation about its observations.
Through Astra, Gemini 2 can search the web for information pertinent to a user’s environment and utilize Google Lens and Maps. Additionally, it can remember what it has observed and heard—though Google states that users can delete this data—enabling it to learn a user’s preferences and interests.
Comments