For most people, researching a subject for school or work-related duties -- or simply learning more about a given topic -- typically entails poring over numerous articles, research papers or tutorial videos. But Sameer Samat, Google's president of the Android ecosystem, tried a different approach recently: Having 30- to 40-minute conversations with Gemini, Google's AI-powered virtual assistant.
"I find it really helpful for learning and kind of being able to engage and ask dumb questions without feeling like I'm being judged," Samat told me just ahead of the search giant's Aug. 13 Made by Google event, where the company announced its new Pixel 9 phones on which Gemini is heavily incorporated.
Samat isn't implying Gemini is a replacement for traditional research. Anyone who has followed the rollout of certain Google AI features, like the AI Overviews summaries in Search, knows they've been far from perfect. And Samat, being a Google executive, is of course quick to point out how helpful the tech giant's digital assistant can be.
Read more: We Asked Google's Co-Founder About AI Smart Glasses. Here's What He Said.
But Samat's story underscores an important point about the current era of digital assistants and where they're going. The rise of generative AI, or AI models that can generate content and provide conversational responses to prompts, has resulted in a virtual assistant renaissance. And Google is determined to show that Gemini isn't just the Google Assistant 2.0 but an entirely new type of helper that can hold lengthy conversations, understand context and interpret multiple types of inputs like sights, sounds and text.
Getting this right is critical for Google because generative AI is believed to be the next major evolution in how we access information and interact with the internet, and every tech giant from Microsoft to Apple, Samsung and OpenAI is infusing the technology into their major products.
"We really think it's a time to reimagine what phones can do," Samat said. "And we think about it as rebuilding Android with AI at the core."
Google's new Pixel 9 phones provide a taste of how the company believes generative artificial intelligence will make our smartphone software more, well, intelligent. The phones come with new apps for extracting relevant information from screenshots and generating custom images on the fly based on prompts. They can also take notes during your phone calls and provide weather summaries. Google's Gemini assistant, which is available broadly across Android and isn't just for Pixel devices, will also work with apps more easily. For example, you'll be able to generate an image with Gemini and drag and drop it right into an email.
But according to Sissie Hsiao, Google's vice president and general manager of Gemini experiences, Gemini's potential stretches far beyond these simple tasks. It's not just about answering questions and retrieving information, much like the versions of Amazon's Alexa, the Google Assistant and Apple's Siri that defined the past decade of virtual assistants. Instead, Gemini will be able to actually complete tasks on your behalf based on your calendar, email and other factors.
Google teased this idea at Google I/O in May, when CEO Sundar Pichai provided an example of a futuristic virtual agent that can even handle real-world tasks like returning a pair of shoes on your behalf by cross referencing information from your email inbox. Google isn't alone; Apple is planning to bring big upgrades to Siri that can provide more personalized answers based on context from your phone.
"This is not your traditional assistant anymore," Hsiao said in a joint interview with Samat before the event. "This is really capable of doing new things."
Instead of just using a helper like the Google Assistant to set a timer while cooking, for example, Hsiao sees Gemini being involved in the whole meal preparation and planning process. That means using Gemini to figure out what to make, what items to add to your grocery list and create invitations to your dinner party and distribute them to your guests. In other words, a lot of the tasks you might traditionally use Google Search for, and then some.
"The whole journey becomes AI-helped," she said.
Gemini Live, the more conversational version of Gemini for paid subscribers that sounds more natural and can be interrupted with questions and clarifications, is a big part of that vision. But Google has more ambitious plans that aren't very far off. At Google I/O, the company showcased a prototype for a new type of virtual assistant called Project Astra, which can "see" your environment using your phone's camera and answer questions accordingly.
Watch this: I Tried Google's Project Astra
04:21
The demo video from Google's conference showed a user pointing the camera at a line of code and asking what it does, for example. The video also briefly teased how this type of assistant could work in a pair of smart glasses, although Google hasn't said anything about whether the glasses shown in the video would result in an upcoming product.
But Hsiao did say the type of functionality shown in the Project Astra video will come to Gemini on smartphones, although she didn't provide a time frame since there's more research and work to be done when it comes to privacy and safety development. She also said Google's phone-based Gemini assistant is the "first" place you'll be able to use Astra.
"We're being bold and responsible with applying that kind of new technology," she said in reference to Project Astra. "But we're super excited that one of the first places that Astra will come to life is in Gemini on the phone."
Google's spotlight on Gemini raises questions about the future of the Google Assistant, which was once a key part of core products from the company's Pixel phones to its Nest smart speakers. Gemini is taking over as the default assistant on the Pixel 9 family, although users will still have the option to use the Google Assistant instead. Hsiao says Google is pursuing two different directions for Gemini and the Google Assistant: Gemini is meant to be more personal, which is why it lives on your phone, while the Google Assistant will be geared toward more communal use on shared devices, like smart home gadgets.
In fact, Gemini's future involves adapting to you and learning about your interests and preferences. She referenced Samat's earlier example, noting that his version of Gemini might be more well-versed in providing answers about technology-related topics based on his preferences and the types of questions he's asked in the past.
"It will get more refined to be Sameer's assistant as opposed to the same assistant for everybody," she said as an example. "And that's a very exciting space [that] I think will be cracked in the next two years for sure."