🔍 Google I/O 2024: Conference Where Google Tried to Catch Up with OpenAI
Google held its annual conference, which was almost entirely dedicated to AI. Despite facing challenges — the event took place right after the cutting-edge GPT-4o presentation — Google's new offerings proved to be impressive.
⚡️ Powerful Gemini
Google introduced a new model: Gemini 1.5 Flash. This multimodal model is optimized for "narrow, high-frequency tasks with low latency." It allows for a better generation of quick responses. However, the latency compared to the streaming conversation with GPT-4o is rather disappointing. The context window of Gemini 1.5 Pro has been increased to 2 million tokens, which is 16 times more than GPT-4o.
Additionally, Gemini will be integrated into the Google Search system to provide direct answers to search queries, similar to Perplexity.
🔍 Video Search
You can record a video of what you want to find and ask a question, and the AI assistant will try to get answers from the internet.
👀 Video Generation
To catch up with Sora, Google is releasing the Veo model, which can create 1080p videos in various styles based on textual descriptions, such as slow-motion mode. The release date remains unknown, but you can join the waiting list.
💬 Personal Assistant Astra
Astra is a multimodal AI assistant that can see and understand what it sees through your device's camera, remember where your things are, and perform tasks for you.
📌 Other Products:
→ Imagen 3 — an AI model for generating images, including the ability to render text.
→ Gems — an application for creating bots (similar to GPTs).
→ Gemini Live — this feature makes voice chats more natural. Users will be able to interrupt the AI assistant mid-sentence.
→ Circle — helps solve mathematical problems.