A Google Search executive said one of the company’s biggest opportunities in AI lies in its ability to know users better and personalize their responses.
AI promises to be uniquely helpful because it knows you. But the risk is that AI feels more like surveillance than a service.
On a recent episode of the Limitless podcast, Robbie Stein, vice president of product for Google Search, explained that Google’s AI tends to process more queries that ask for advice or where users are looking for recommendations, and that these types of questions are likely to benefit from more subjective responses.
“We think there’s a huge opportunity for our AI to get to know you better and be uniquely helpful with that knowledge,” Stein said in an interview. “And one of the things we talked about there was [Google’s developer conference] I/O was a way for AI to better understand users through connected services like Gmail. ”
Google has been integrating AI into its apps for some time, ever since Gemini was still known as Bard. We recently started ingesting personal data into another AI product, Gemini Deep Research. And now, Gemini is built into Google Workspace apps like Gmail, Calendar, and Drive.
But as Google integrates more personal data into its AI, including emails, documents, photos, location history, browsing behavior, and more, the line between helpful and annoying assistants is becoming increasingly blurred. And unlike opt-in services, as AI becomes more central to Google’s products, it may become harder to avoid data collection by Google.
Google’s pitch is that this deep personalization will make AI even more useful. The idea is that Google’s AI technology can learn from your interactions across Google’s various services and use that understanding to make more personalized recommendations. For example, if we learn that users prefer certain products or brands, the AI response may prioritize those products or brands in its recommendations.
Stein said that’s “much more convenient” than showing users a more general list of best-selling products in a particular category. “I think that’s really the vision of building something that’s really knowledgeable to me.”
This idea is not all that different from how the “other” in Apple TV’s hit show “Pluribus” devoured knowledge of the world, including the intimate details of individuals. When the system interacts with the show’s main character, Carol, it uses that data to personalize everything. That means cooking her favorite meals, employing familiar faces to handle communications with her, and otherwise anticipating her needs.
But Carol doesn’t think personalized responses are kind. She finds them invasive. She never agreed to share her data with the hivemind, but the hivemind knows more about her than she would like.
Similarly, it seems increasingly difficult to circumvent Google’s data-cannibalization techniques in the AI era, and unless Google strikes the right balance, the results could feel more creepy than helpful.
(Just to be clear: Google lets you control which apps Gemini’s AI uses to learn more about you. This can be found under “Connected Apps” in Gemini’s settings.)
If you share app data with Gemini, Google says it will store and use that data in accordance with Gemini’s privacy policy. And the policy warns users not to “enter sensitive information that you don’t want reviewers to see or that you don’t want us to use to improve our services,” as human reviewers may be able to read some of your data.
But as more data enters Google’s own collective consciousness, it’s easy to see how AI could make data privacy more of a gray area.
But Google believes it has some kind of solution.
Stein said it will indicate when Google’s AI responses will become personalized.
“I think people want to intuitively understand when they are personalized, when information is being created for them and when it is being created for them. [it’s] “Anyone who asks this question will understand,” he said.
Stein also pointed out that Google could send push notifications to users when a product they were considering after days of online research becomes available or launched.
“Today, Google can [is] “And I think that’s how I think about the future of search, more than any particular feature or single form factor,” he said.
