Gemini's Personal Intelligence: Helpful, Error-Prone, Intrusive

Google’s Gemini now offers a feature called Personal Intelligence that — if you opt in — lets the chatbot reference past conversations and pull from your Gmail, Calendar, Photos, and search history without you asking it to check those sources. The capability is in beta and limited to people on Google’s AI Pro and Ultra tiers. Previously Gemini could access Workspace apps, but typically only when explicitly prompted; Personal Intelligence makes those connections more proactive.

In hands-on tests, Gemini’s suggestions can be surprisingly on target: recommended books matched the author’s tastes, and a planning prompt produced a lawn-care plan with native-plant ideas, calendar reminders, and a Keep shopping list. The assistant has also improved at completing actions it used to fail at, like adding events to a calendar.

But the article stresses practical limits. Gemini often trips on details — provided map routes that didn’t match the linked directions, suggested businesses in the wrong neighborhoods, and recommended a shop that was clearly closed. Those kinds of errors can make a supposedly helpful assistant feel like more work than it saves.

There are privacy concerns as well: because Gemini can surface names and personal details from email and calendar data, that can feel intrusive even if access is opt-in. The author says Personal Intelligence broadens Gemini’s utility only modestly given the lingering accuracy and privacy issues.