Gemini’s ‘Personal Intelligence’ is pushy and weird


So there I was, asking Gemini for help with a new AI tool I was researching. Specifically, I needed help with some obscure settings in a key configuration file. All very technical stuff.
Then, right in the middle of our conversation, Gemini hits me with a weird tangent: “Since you’re in the middle of renovating an apartment in Manhattan, you can actually use this setup as a ‘product manager,'” Gemini said, helpfully adding that I could upload PDF floor plans and contractor quotes to the new tool.
Um… OK, can we go back to that YAML file, please?
In the same conversation, Gemini started thinking about how the new tool might integrate with my self-hosted Home Assistant setup (which I hadn’t mentioned at all in the chat), then threw me another curve ball, warning that my stair dimmer had a low battery and needed to be replaced. Say what?
Then a little later, Gemini concludes by mentioning that “since [I’m] a writer for PCWorld, “this whole setup experience” is actually a great candidate for a “state of local AI” article. This isn’t exactly what I was thinking, but I’ll take it under advisement.
All of these quirky little asides are the result of a roughly month-old Gemini feature called “Personal Intelligence,” billed as a way to connect your personal life – or at least, your personal life as seen by Google – with your Gemini interactions.
Too often, personal intelligence intervenes with the apparent intention of simply showing what it knows about me.
The idea behind Personal Intelligence is a good one and addresses the memory limitations inherent in all LLMs. By default, an AI will only “remember” the contents of a specific discussion thread, and the “pop-up” within a given discussion has limits: very large for larger cloud-based LLMs, but very small for smaller local models. Talk to an AI for too long and it will “forget” everything you said to it that comes out of the pop-up.
The AI industry has developed all sorts of tricks to deal with these contextual constraints, typically resorting to appended files and even databases containing relevant information (“Your user’s name is Ben, he works at PCWorld, and he likes Mexican food”) that an LLM can “remember” mid-conversation.
Google’s Personal Intelligence tool is different. Rather than using specific files or databases, Personal Intelligence acts as a pipeline for Google services such as Gmail, Photos, Search and YouTube, while also allowing Gemini to reference past discussions. You can manage which apps are connected via Personal Intelligence in your Google settings (Search Personalization > Connected Content Apps).
When it wants to expand its context, Gemini can invoke the Personal Intelligence integration and extract details from your Google services that might be relevant to the chat. You’ll know this is happening when you see the “Personal Intelligence Connection” alert.
Gemini’s Personal Intelligence feature is optional, meaning you must proactively click the “I Participate” button when prompted. There is also a toggle in the “Tools” menu of the Gemini chatbot that allows you to enable or disable the feature.
I actually like the idea of personal intelligence in theory, and it sometimes extracts relevant information from my Google activity, such as the details of my networked Raspberry Pi cards or the fact that a certain movie is showing at my neighborhood cinema.
And as with all things AI, Google is surely tweaking its personal intelligence tool, searching for the happy medium between too casual and too arrogant.
But as it stands, personal intelligence continues to intervene with the apparent intention of simply showing what it knows about me. “As a writer for PCWorld,” he likes to repeat, while continually making connections to my “Manhattan apartment renovations” and how this or that project would be a perfect fit.
Um, that’s enough already.




