Google on Tuesday highlighted its commitment to artificial intelligence (AI) by revealing a range of new AI products at its annual I/O developer conference.
The announcements, including new AI hardware made specifically for cloud users, follow competitors' recent AI breakthroughs.
Earlier this month, Amazon-backed Anthropic unveiled its first enterprise product, while OpenAI debuted a new AI model and a desktop version of ChatGPT.
AI-Powered Improvements and Add-Ons
The tech giant has updated its Gemini AI model, which now comprises the Gemini 1.5 Pro and Gemini 1.5 Flash. While the Flash model is more affordable and intended for minor tasks like summarizing conversations, captioning movies, and extracting data from documents, the Pro version can handle large amounts of data, such as summarizing 1,500 pages of text.
With its expanded translation skills to 35 languages, Gemini is a formidable rival to Apple's Siri and will eventually take the place of Google Assistant on Android phones.
Creative AI Generative Models
New generative AI models have also been showcased during the conference, including Imagen 3 for text-to-image generation and Google Veo for high-definition video production that produces lifelike visuals with fewer visual artefacts. Selected developers will have access to these tools, which will also be incorporated into Google's Vertex AI platform, which is used to train and run AI apps.
With interactive audio explanations, the "Audio Overviews" function can also produce audio summaries from text inputs, improving the user experience.
Improved Search Features
Google Search is about to undergo a major overhaul with the launch of "AI Overviews", which offer concise explanations of intricate search terms. This innovation, which goes live in the United States on Monday, will provide summarized data, such as multi-step task procedures, at the top of search results. Future upgrades will let users ask inquiries via photos or videos and provide multimodal search choices with assistant-like planning skills.
AI Incorporation into Google Workspace
Users of Google Workspace will be able to manage and search through work papers and email threads with the help of a new feature called "AI Teammate", which will integrate AI. This helper offers thorough support for project management and decision-making by summarizing and analysing data from multiple Workspace apps.
The Future AI Assistant, Project Astra
Project Astra, created by Google's DeepMind AI section, seeks to be the company's version of J.A.R.V.I.S., the all-knowing AI assistant from the Marvel Universe. Currently a prototype, Astra may engage in real-time via video and voice, providing seamless conversational assistance. This ambitious initiative is scheduled to be integrated into Gemini later this year.
AI Hardware Advances
Google has revealed Trillium, their sixth-generation TPU (tensor processing unit), which will be available to cloud customers in late 2024. This gear is critical for carrying out complicated AI functions. Google will also provide Nvidia's Blackwell GPUs in early 2025, as part of a long-standing cooperation with Nvidia to improve its AI infrastructure for enterprise developers.