Google’s Gemini AI Chatbot Introduces Memory Function Amid Privacy Concerns

Google’s AI chatbot, Gemini, now has a memory function. This feature allows the AI to store user data and preferences, making responses more personalized and tailored. Google claims that answers with stored memories are “more helpful and relevant.” However, this raises concerns about privacy.

Over a year ago, Google announced plans to add a memory function to its chatbot, then known as Bard, for personalized responses. Until now, Gemini could not remember conversation details, only information like language, device, and location. This is changing as Google catches up with competitors. OpenAI introduced Memory for ChatGPT in early 2024, giving its chatbot a memory.

Gemini can now remember user preferences, such as dinner choices, so restaurant recommendations weeks later can be customized. Users can specify what the chatbot should remember using phrases like “Don’t forget” or “Always say.” According to Google, this feature can be turned off, and memories can be individually deleted by the user.

The memory function is currently available only in English and through the browser. The mobile Gemini apps for Android and iOS do not yet have this capability. Additionally, only “Gemini Advanced” offers this feature, requiring a “Google One AI Premium” subscription. This subscription costs 21.99 euros per month and includes additional cloud storage (2 TB) along with the enhanced AI chatbot. In contrast, ChatGPT’s Memory function has been free for all users since early September.

There is also the question of whether Google uses stored user memories for training its AI language models. Google denied this in response to a TechCrunch inquiry, stating, “Your stored information will never be shared or used to train the model.” This aims to protect user privacy, as nearly a year ago, ChatGPT was tricked into revealing training data through a word repetition trick.

Despite assurances, there is criticism about storing potentially sensitive user data by AI chatbots. Some consider it a privacy disaster, likening OpenAI’s use of the Memory function to personalized advertising, which is seen as equally bad or worse in terms of privacy.

Related topics include updates on AI policies, drone defense, and learning materials. Google has launched its own iPhone app for Gemini AI, and there are discussions on natural conversations with ChatGPT and Gemini. GitHub Copilot now supports Claude and Gemini instead of GPT.

Exit mobile version