Summary
Summary
- Gemini’s new Memory summary displays the broader personal profile it has built from your conversations.
- Users can correct or delete inaccurate details and review when changes were made through Memory history.
- As AI assistants become more personal, transparent and accurate memory could prove as important as raw computing power. Google is rolling out a new Memory summary that finally shows you what Gemini has learned about you through your chats and Google Account activity. According to Google’s official support page, the section lets you inspect, correct, and delete personal details Gemini uses to tailor its responses. The summary covers Gemini’s understanding of your background and preferences and includes many details that have been collected gradually across previous conversations. The important bit is that not only can you analyze the memory, but you can now correct inaccurate assumptions or ask it to forget certain information that you no longer think is needed for future responses. Gemini’s Memory is finally catching up to rivals This has long been one of Gemini’s most conspicuous disadvantages. In a previous experiment, we taught ChatGPT, Claude, and Gemini the same lie and then tried to erase it. ChatGPT and Claude both provided readable summaries you could manually edit. Gemini, on the other hand, left us scratching our heads. While it remembered the false information, it offered no list showing where it was stored. Deleting the relevant conversations and testing it again was the only practical way to determine whether the lie had disappeared. Thankfully, Google is now catching up to its rivals in this regard. ChatGPT already allows you to view your memory summary and allows you to make changes. That said, it’s important to note that these summaries are not a complete record of every piece of data Google holds. Gemini’s Memory, chat history, Connected Apps, and the activity saved to your Google account are related but separate systems. In general, it’s safe to assume that Google knows a lot more about you than the summary it gives you in Gemini. Moreover, bear in mind that editing a remembered detail will not necessarily delete the conversations or other shared data from which Gemini originally inferred it. You can find the new summary under Settings > Personal Intelligence > Memory summary, or simply ask Gemini, “What do you know about me?” You can then make changes through written prompts. You can add a detail, correct an inaccurate entry, or ask it to remove outdated information. You can navigate to the main Personal Intelligence menu in order to disable Memory altogether. For now, Memory summary is limited to US users with an active Google AI plan and only works in English. It is available through the Gemini mobile and web apps, so long as both Keep Activity and Memory are enabled. Which is more important: memory context or computing power? AI companies usually compete by showing benchmark scores, context windows, and claims about which model can solve more complex problems. For a personal assistant, however, raw intelligence loses much of its advantage if you have to begin every conversation by explaining the same details of an ongoing project. Memory plays an important role in the daily use of chatbots like Gemini and ChatGPT. A slightly weaker model that understands your routines better can produce a much better answer than a capable one working without context. Gemini Personal Intelligence, for instance, can combine past chats with information from services such as Gmail, Drive, Photos, Search, and YouTube to produce individualized responses. You don’t need Anthropic’s recent Fable 5.2 to tackle such tasks. For better or worse, AIs are getting more deeply embedded in people’s work and private lives. Given how frequently users interact with them, the quality of their memory (and the user’s ability to audit it and make necessary changes) becomes just as important as the model underneath. Gemini’s new summary gives users a long-overdue look inside that process.