Gemini

Google Cloud is adding generative AI to its cloud AI portfolio. Vertex AI will provide foundation models (initially text and images, later audio/video) with tools to discover models, edit prompts, fine-tune on customer data, and deploy. Google also launched Generative AI App Builder to create chat interfaces and digital assistants that integrate search and foundation models. New partner programs expand the AI ecosystem; trusted testers have access starting today.

Google is providing limited access to Bard, a conversational AI based on LaMDA, initially available to select users in the US and UK. Access is being managed via a waitlist at bard.google.com. Bard derives from Google’s LaMDA work begun in 2015 and was trained on roughly 1.56 trillion words from public forums and other websites.

Google released Gemini, its most capable large language model, first announced at Google I/O. Starting today, a specially tuned English-only version—Gemini Pro—is available in Google Bard, bringing the new model’s capabilities directly to Bard users. Sources: Google, ZDNet.

Google released Gemini Ultra 1.0, its largest and most capable large language model, now available to the public. Google is retiring the Bard name and consolidating Workspace AI (formerly Duet AI) under the Gemini brand. On Android, the Gemini app can be set as the default assistant, replacing Google Assistant.

Tianyu Xu published a “Gemini Advanced Guide” covering image creation, writing, research and analytics, and integration with Google. The guide includes 35 ready-to-use prompts and screenshots illustrating workflows and outputs for Gemini Advanced users.