Kerry Wan/ZDNETGoogle has had a busy year so far, with the rebranding of its AI chatbot from Bard to Gemini and the launch of several new AI models. At the recent Google I/O developer conference, the company made a series of announcements regarding AI and its integration into various apps and services.
AI was the main focus of the event, with Google incorporating the technology into nearly all of its products, including Search, Android 15, and Gemini. Here’s a summary of the major announcements from the event:
1. Gemini:
Google introduced Gemini 1.5 Flash, the fastest Gemini model available in the API, offering a more cost-effective option than Gemini 1.5 Pro. Gemini 1.5 Flash is now in public preview in Google’s AI studio and Vertex AI. Additionally, Gemini 1.5 Pro has been upgraded to provide better responses in various areas, including translation, reasoning, and coding.
Gemini 1.5 Pro now offers a 1 million context window for consumers in Gemini Advanced, allowing for AI assistance on large bodies of work. Google is also previewing a two million context window for developers in Gemini 1.5 Pro and Gemini 1.5 Flash.
Gemini Nano, designed for smartphones, now supports images in addition to text. The Gemma family of models is also getting an upgrade with the launch of Gemma 2 in June, optimized for TPUs and GPUs. PaliGemma, Google’s first vision-language model, is joining the Gemma family.
2. Google Search:
The AI overview feature, previously available in Search Labs, is now accessible to all users in the U.S. The feature, powered by a new Gemini model customized for Google Search, provides conversational answers to search queries. AI-organized search results and new features like video search, meal, and trip planning have also been introduced.
3. Google Assistant:
Generative AI is a key focus at Google I/O, with over 10 developer sessions dedicated to topics related to generative AI, including Gemma advancements and multimodal retrieval-augmented generation with Gemini.
4. Veo (text-to-video generator):
Google unveiled Veo, a text-to-video model capable of generating high-quality videos beyond a minute in length. The model can understand natural language and cinematic terms to create videos that align with user vision.
5. Imagen 3 (text-to-image generator):
Google introduced Imagen 3, a next-generation text-to-image generator that produces high-quality images with improved natural language capabilities.
6. SynthID updates:
Google expanded SynthID, its AI-labeling tool, to include text and video modalities. The tool will watermark videos generated by Veo.
7. Ask Photos:
Google introduced Ask Photos, allowing users to use conversational prompts in Google Photos to find specific images.
8. Gemini Advanced upgrades (featuring Gemini Live):
Google upgraded its Gemini Advanced subscription tier, offering users access to Gemini 1.5 Pro and unique experiences. Gemini Live, a real-time voice AI bot, is also part of the upgraded offerings.