Google has unveiled a memory feature for its Gemini AI platform, allowing users to teach the system their preferences and priorities. The new feature is exclusive to Gemini Advanced subscribers under the Google One AI Premium Plan, enabling a more customized AI experience.
For instance, users can instruct the AI to “Remember I’m vegetarian” or “Don’t forget I prefer brief responses,” and these details persist across interactions.
Information saved by Gemini can be managed through a dedicated “Saved Info” page on its website, where users can view, edit, or delete entries. Currently available only in English, this feature gives users more control over their AI while reducing the need to repeat preferences repeatedly.
Gemini-Exp-1114 Surpasses GPT-4o in AI Performance
While Google is adding features, a new experimental model called Gemini-Exp-1114 has claimed the top spot in the Chatbot Arena leaderboard, a competitive benchmark where AI systems are tested blindly by users across tasks like problem-solving and dialogue management.
With over 6,000 user votes, Gemini-Exp-1114 achieved a score of 1344, outperforming OpenAI’s GPT-4o, which had led since September.
The model excels in advanced reasoning, such as solving mathematical problems and engaging in multi-turn conversations. However, it remains accessible only through Google AI Studio, a platform designed for developers experimenting with emerging technologies.
| Related: |
Enhancing Productivity: Gemini AI in Google Workspace
Google´s recent Gemini updates brought deeper integration into Google Workspace tools like Docs and Gmail. In Docs, Gemini now powers image generation through the Imagen 3 model, allowing users to create visuals by providing simple prompts. Meanwhile, Gmail users benefit from integration with Google Calendar, enabling scheduling tasks without leaving their inbox.
Although these tools simplify workflows, they currently lack advanced features like managing guest lists or updating visuals dynamically. Nevertheless, these additions demonstrate Google’s commitment to embedding AI across its ecosystem.
Gemini Live Comes to iPhone
Expanding its presence on mobile, Google launched a standalone app of Gemini AI for iPhone earlier this month. This app supports voice commands and interactive conversations through Gemini Live, which offers ten distinct voice profiles, such as “Ursa” with a mid-range tone and “Capella,” known for its British accent.
Recent beta testing suggests that Gemini Live will soon enable voice-driven discussions about uploaded files, a feature likely to improve mobile productivity significantly.
| Related: |
Customized AI Chatbots with Dynamic Document Integration
For business and educational users, a new Gemini’s Gems feature introduced the option to use uploaded documents for better responses. Users can add up to 10 files, such as PDFs or spreadsheets, and benefit from automatic updates when these files are stored in Google Drive.
This seamless integration contrasts with OpenAI’s Custom GPTs, which require manual uploads and lack real-time syncing. While Gems cannot be shared publicly, they offer a secure and efficient way to automate repetitive tasks.
Google’s Gemini advancements arrive amidst growing competition in the AI sector. OpenAI’s Orion model is reportedly facing delays due to compute limitations and reliance on synthetic data, while Anthropic and other contenders like Mistral aim to close the intelligence gap and beating OpenAI with better features.
