29. 05. 2024 | Natalie Bezděková

0

• Share with others    

The Google I/O developer conference, held on May 14, brought a number of significant innovations with an emphasis on advances in artificial intelligence. Sundar Pichai, Google’s CEO, introduced several key innovations.

Google introduced a major enhancement to its AI model, Gemini 1.5 Pro. This model, originally launched with an information processing capacity of 1 million tokens, now has an expanded capacity to 2 million tokens. This expansion will enable the Gemini 1.5 Pro to better understand complex queries and provide more accurate answers. In addition, Google introduced Gemini Flash, a more affordable AI model optimized for smaller and more specific natural language processing (NLP) tasks.

Gmail also saw significant improvements. Gemini’s model integration allows attachments to be analyzed directly in emails, making it easier to navigate long threads and quickly find key information through automatic summaries.

Google also introduced Veo, a new video generation model that competes with OpenAI’s Sora. Veo can create high-quality 1080p videos based on text descriptions, taking the possibilities of AI-powered video creation to a new level.

Another new feature is Audio Overview, a feature for automatically generating audio summaries from text documents. This innovation makes it easier to review document content and is available in the US for now.

AI Sandbox, another featured tool, uses generative AI to create music and sounds from the ground up. Users can experiment with music creation using simple text-based assignments that generate sounds, melodies and rhythms.

Google search has also been improved with AI Overviews, which provides concise and to-the-point answers to complex queries. Multimodal search, which allows queries to be asked using video, is another major new feature. This allows Google to analyze video content and provide relevant results.

The ambitious Project Astra represents progress in the development of advanced AI assistants. Capable of processing text, video, image and audio, this versatile agent promises to revolutionise the everyday use of AI.

Google also announced an update to its open-source models. The next generation of Gemma 2 brings improvements in efficiency and scalability, allowing models to run on a single TPU or GPU. Gemma 2 thus makes advanced AI models available to a wider developer community.

The Google I/O 2023 developer conference brought many exciting new developments that push the boundaries of AI and bring new possibilities for developers and end users alike.

Natalie Bezděková

I am a student of Master's degree in Political Science. I am interested in marketing, especially copywriting and social media. I also focus on political and social events at home and abroad and technological innovations. My free time is filled with sports, reading and a passion for travel.