Google's Gemini AI Can Now Remember Things You Tell It
Google has introduced a new memory feature for Gemini, their AI technology. This feature enables Gemini to retain information shared with it, allowing the AI to incorporate user preferences into its future interactions. The initial discovery of this capability was made by TestingCatalog News. Subsequently, Google officially announced the feature through a video demonstration showcasing its functionality.
One demonstration featured Gemini being instructed to consistently suggest two or three subject lines while composing an email. In a follow-up scenario depicted in the video, Gemini generated an email containing three subject line options. Google emphasized the value of this feature in a blog post, underlining its ability to enhance efficiency and creativity by organizing personal and professional data.
Personalized Responses
According to 9to5Google, the memory feature operates as expected. Users can inform Gemini about themselves and their preferences. Gemini utilizes this information to tailor responses accordingly. Examples include specifying dietary restrictions or requesting code examples in a particular programming language. Users can further customize Gemini's behavior by enabling language translations, controlling technical terminology usage, or managing response lengths.
At present, the memory feature is exclusively accessible in English to subscribers of the $20-per-month Gemini Advanced service.
Limitations and Comparisons
9to5Google reports that this release is confined to the web version of Gemini and is not yet compatible with the mobile apps. This development places Gemini on par with its major competitor, ChatGPT, which also incorporated memory features earlier this year. ChatGPT's memory feature is accessible to all users, in contrast to Gemini's subscription-based approach.
Unlike Gemini Advanced, where the memory feature is limited to paid subscribers, ChatGPT offers its memory function to free users. Both AI technologies enable users to input information for personalized responses in subsequent interactions.
