Major AI Announcements at Google I/O 2024: From Gemini Upgrades to Project Astra

Google made several announcements regarding artificial intelligence (AI) during its annual I/O 2024 developer conference last week. The company detailed how it’s integrating AI across its various apps and services, including its multimodal AI tool, Gemini.

During the two-hour conference, Google demoed its new AI assistant, codenamed “Project Astra.” The assistant is termed “a universal AI agent that is helpful in everyday life.” The AI assistant, with its multimodal functionality and ability to take action in real-time, works on smartphones and smart glasses.

“Google is fully in our Gemini era. We’ve been investing in AI for more than a decade — and innovating at every layer of the stack: research, product, infrastructure,” Google CEO Sundar Pichai said in the introduction to his keynote speech.

New AI Upgrades: Gemini Dominates

The conference highlight was the new AI upgrades to Google’s flagship products, including search, workspace, and Gemini 1.5 models.
The company introduced the new Gemini 1.5 Flash, which is expected to generate quicker responses and be more cost-effective to deploy. The new model is currently available in public preview in Google’s AI studio and Vertex AI.

Another notable announcement was the upcoming “Gemini Live,” which enables users to engage in a two-way conversation with Gemini and even interrupt it mid-conversation.

Google is also infusing AI into its Workspace. The Gemini side panel of apps like Gmail, Slides, Docs, and Sheets will be upgraded to implement Gemini 1.5 Pro.

Gemini 1.5 Pro also received an upgrade—with a context window of up to 2 million tokens. The upgrade means that it can now answer questions based on larger texts, such as thousands of pages or long-form videos.

Gemini 1.5 Pro, with its 1 million context window, is available in 35 languages to Gemini Advanced subscribers.

Search Evolves with AI

The new AI Overview feature pools information from various web sources to answer search box queries. Users no longer need to visit specific pages while searching for answers. AI Overview is already available to users in the US and is expected to be available in other countries soon.

Search also includes a Multi-Step Reasoning feature. This helps break down complex searches into an organized step-by-step solution. For example, during the live demo, the feature was used to plan a trip, and it provided the user with a detailed itinerary, including meal plans.

Some other capabilities demonstrated were the soon-to-be-available AI-organized results page and a video search feature.

Veo: A New AI Creativity Tool

Google introduced its new multimodal, text-to-video generator tool, Veo. The creativity software can produce high-quality 1080p videos that are more than a minute long. It can also understand cinematic terms like aerial shots and generate videos in various styles.

Google demoed the model’s functionality via a video featuring filmmaker, actor, and musician Donald Glover sharing his experience experimenting with the tool. Veo is now available to a select group of creators in Google Lab.

It also announced the next-generation text-to-image generator, Imagen 3, which, according to Google, helps create more realistic pictures.

Reimagining Android With AI

Google announced several AI features that would be built directly into its Android ecosystem like the already available Circle to Search feature and the replacement of Google Assistant with Gemini. The opt-in AI assistant, with its multimodal capability, will be overlayed across various apps and will provide answers when requested.

Gemini Nano is also being upgraded with multimodal capabilities and will be integrated with Google’s TalkBack accessibility feature. It will provide more descriptive information for users with blindness or low vision. Gemini Nano will also function as an opt-in scam detector, listening to and detecting suspicious conversation patterns to warn users in real-time. It will be available later this year.