Google Assistant Gets a Gemini Upgrade

Google Assistant, as we’ve known it, is evolving. The core technology that powers Google’s virtual assistant is undergoing a significant change. Late last year, Google integrated AI capabilities from its chatbot, Bard, into Google Assistant. This integration allowed the assistant to interpret images and extract data from emails, documents, and more. Now, Google is offering users the option to replace the AI models that drive Google Assistant’s conversational abilities with Gemini, Google’s latest GenAI technology.

In the near future, Android users will be able to access a new Gemini-powered Google Assistant experience through a new Gemini app. This app will replace the existing Google Assistant app. iOS users will also have access to this new experience through an updated Google app, set to be released in the coming weeks. However, this upgrade will initially be mobile-only. Google Assistant devices like Nest and Home speakers and displays will not see changes immediately.

The new experience is presented as an overlay that serves as a frontend for the Gemini models. It provides contextual recommendations and suggestions. Sissie Hsiao, Google Assistant VP, explained in a press briefing that this overlay enables users to do some really interesting things with Gemini. For instance, users can generate a caption based on a picture they’ve already taken that’s behind the overlay. Users can also bring up the overlay while reading an article to understand the article more deeply and ask questions about it.

The Gemini-powered Assistant can accept images as well as text and voice commands. For example, users can upload a picture of a plant with the instructions “I just bought this plant — give me tips and YouTube videos to help me take care of it,” and the Gemini-powered Assistant will do its best to fulfill the request.

Gemini is capable of a range of GenAI tasks, a fact that users who have tried a Gemini model on the web via Bard are likely aware of. This capability has not changed with the Assistant integration. Google provides a few example prompts in the onboarding flow, such as “Help me craft a text response to my friend who is stressed at work,” “Give me some ideas to surprise my concert-loving friend on their birthday,” and “Help me incorporate more plant-based options in my diet.” Gemini in Assistant can also create images, although it’s not clear which model is doing the generating.

Conversations with the Gemini-powered Assistant on mobile will carry over to Gemini-powered experiences on the web and vice versa, provided users are signed into the same Google Account in both places.

If users are not satisfied with the Gemini-powered Assistant experience, they can switch back to the old models at any time through a menu screen in the Assistant app on Android and the Google app on iOS.

However, the Gemini models are not completely replacing the old models that underpin Google Assistant. Features including quick phrases, which let users skip saying “Hey Google” for specific tasks, will continue using the old models. So will features related to setting timers, making calls, and controlling smart home devices. Google notes that even when the Gemini models are selected as the default models in Assistant, apps with Assistant integration like Google Maps may continue using the old models.

In a move that may disappoint some longtime Google Assistant users, the most capable of the Gemini models, Gemini Ultra, is not available to power Assistant for free. It and the capabilities it brings, such as better reasoning, coding, and instruction-following skills, are part of a new product, Gemini Advanced, which is not free.

What is the difference between Bard and Gemini?

Bard and Gemini are both powerful technologies developed by Google, but they serve different purposes and have distinct capabilities12:

In essence, Gemini and Bard have a relationship similar to fuel and a vehicle. Gemini operates as a powerful force driving different applications, much like how fuel propels various technologies1. On the other hand, Bard functions like a vehicle, drawing its energy from the capabilities provided by Gemini.

What is LaMDA?


LaMDA, short for “Language Model for Dialogue Applications”, is a research breakthrough by Google1. It’s designed to improve the versatility and adaptability of language models, making them more capable of handling open-ended conversations1.

Unlike most other language models, LaMDA was trained on dialogue, allowing it to pick up on several nuances that distinguish open-ended conversation from other forms of language1. For instance, it can understand the sensibleness of a response in a given conversational context.

LaMDA can engage in a free-flowing way about a seemingly endless number of topics, which could unlock more natural ways of interacting with technology and entirely new categories of helpful applications. It’s built on Transformer, a neural network architecture that Google Research invented and open-sourced in 2017.

However, it’s important to note that LaMDA is different from AWS Lambda, which is a computing service that lets you run code without provisioning or managing servers2. The term “lambda” is also used in computer science to refer to anonymous functions, and in mathematics and physics to represent various concepts.

LaMDA and GPT-4 are both advanced language models developed by Google and OpenAI respectively, but they have different focuses and strengths123:

In terms of availability and pricing, LaMDA has limited availability through AI Test Kitchen, while GPT-4 is available through a monthly subscription costing $201.

Sharing Is Caring:

Leave a Comment