Search AI Products and News

Google has announced the reactivation of the character image generation feature for its AI chatbot Gemini, aimed at producing diverse content including photo-realistic landscapes and textured paintings. This feature is powered by Google's AI text-to-image generator Imagen3 and currently supports only English. It will be available to Gemini Advanced, Business, and Enterprise users in the coming days. The newly upgraded Imagen3 model offers improved performance and includes built-in safety measures optimized to avoid historical inaccuracies.

Google is testing the 'Ask Photos' feature, which allows users to interact with their personal photo library using natural language, such as asking to view the best photos taken at every national park or querying license plate numbers. This feature hasn't been officially released yet but is included in the latest version of the Google app. Through 'Ask Photos', photos and videos will transform into a visual database that can intelligently extract information and context. Reports indicate that keywords found in the code suggest new extended functionalities related to Google Gemini.

Google is leveraging its dominance in the search engine market to force publishers to choose between allowing their content to be used for AI responses or risking a drop in search rankings. Publishers can prevent their content from being used by AI systems through the use of the 'nosnippet' tag, but this move may affect their overall search engine visibility. Google has stopped negotiating with publishers regarding content usage rights, while some AI companies like OpenAI and Perplexity are attempting to address this issue through compensation schemes. In contrast, Apple is also considering revenue through advertising.

Google has introduced a 'refinement' feature to Gmail, aimed at enhancing email quality and ensuring more appropriate expression. This feature is implemented through the Gemini writing tool, allowing users to choose options like 'Formalize', 'Expand', or 'Refine' under the 'Help Me Write' section to optimize email content. The AI writing tool is more intuitive and user-friendly on Android and iOS devices. Users with a Google One AI Premium account or Gmail Workspace can enjoy this feature for quick optimization of new drafts.

Google has launched the Pixel Screenshots app for the Pixel 9 series, aimed at intelligently managing screenshot information. The app uses the Gemini Nano model to analyze image content directly on the device, making screenshots searchable; users can quickly find specific screenshots by keywords, improving information retrieval efficiency. The app supports the creation of custom collections and automatically filters screenshots related to keywords, featuring information extraction capabilities, such as extracting T-shirt prices from images. Currently, it is only available for Pixel 9 series devices.

Google DeepMind's AI systems AlphaProof and AlphaGeometry2 achieved remarkable results in the International Mathematical Olympiad, winning silver medals and demonstrating astonishing capabilities in solving mathematical problems. AlphaGeometry2 solved a geometric problem in 19 seconds, surpassing the speed and accuracy of top human mathematicians. This achievement has been highly praised by legendary figures in the mathematics community, who believe it expands the boundaries of AI applications in the field of mathematics. Terence Tao noted that advances in AI will aid mathematical research.

Google researchers have introduced a pre-trained scoring model called Cappy, aimed at improving the performance of large multitask language models. Cappy's architecture is based on RoBERTa and is pre-trained on a diverse collection of datasets to improve the performance and efficiency of multitask law master's models. By introducing the lightweight pre-trained scoring model Cappy, it addresses the challenge of effectively utilizing large language models in multitask scenarios. Researchers have proposed a data construction method that meets the diversity requirements of labels in pre-training data, generating a large and effective regression pre-training dataset.

Google will hold the 2024 I/O Developer Conference online on May 14, with the keynote speech streamed over the internet to attract a broader developer audience. Google will showcase new features of the Gemini model and Android 15, including upgrades to Gmail and Google Photos, as well as a potential mid-range phone launch. The Gemini model will be embedded in devices to compete with OpenAI's Sora, demonstrating powerful multimodal mixture of experts model capabilities. Google releases Gemini 1.5.

Google is enhancing its ranking system to eliminate AI-generated spam content. This initiative aims to reduce the impact of websites created for search engines rather than for people, expecting to decrease 40% of low-quality search results that lack original content. Google's goal is to direct traffic to high-quality, helpful websites.

Android Auto update introduces a new 'P' badge to remind users to use Android Auto while parked. Version 11.4 introduces an AI summary feature that automatically summarizes long messages or busy group chats and suggests relevant replies and actions.

Google co-founder Sergey Brin made a rare public appearance, where he sharply criticized Google's Gemini image generator. He pointed out existing issues with the Gemini image generator, leading Google to halt its functionalities. Brin stated that the problems faced by the Gemini image generator were not intentional. He emphasized his cautious view on the hallucination issues with large models and expressed strong confidence in Google's business model while supporting the development of artificial general intelligence (AGI). Additionally, Brin commented on Google Glass and mentioned Apple's Vision Pro and Meta's Quest.

CEO Sundar Pichai criticized the image generation errors of Google Gemini AI, calling them 'completely unacceptable.' The team is correcting the issues and emphasizes the importance of providing fair and accurate information, with structural reforms underway. The article was published on 2024-02-28.

Google has released the open source AI large model Gemma, with the domestic open source large model InternLM2 outperforming Gemma in all aspects. The competition in the open source large model field is intense, making the large model landscape increasingly lively.

Gemini, Google's largest and most powerful multimodal AI model, has encountered issues with its character image generation feature, raising concerns about biases in the images it generates. Alphabet has paused this character image generation feature, with plans to restore it in the 'coming weeks.' Google CEO Sundar Pichai stated that the errors in Gemini are 'unacceptable' and proposed solutions and improvements.

Google has partnered with several publishers to create news content using generative AI tools. The collaboration with publishers is valued at tens of thousands of dollars annually and is part of Google's News Initiative. Generative AI tools can help publishers more effectively create aggregated content. Participating publishers are required to publish at least three articles daily. AI-generated text will be represented with a color-coding system to indicate credibility.

Google has released the base world model Genie, which features 11 billion parameters and can generate interactive virtual environments. Genie trains without action labels, autonomously identifying different action features and patterns in videos. Genie can learn fine-grained control and simulate a variety of potential actions. It generates video in a cartoon style and can simulate robotic movements and deformable objects, leaning more towards animated graphics compared to Sora.

Google's Gemini model portrait generation feature is set to relaunch. The portrait generation feature of Gemini did not work as expected, providing inaccurate information in historical image generation. Google is working on fixing the issues and plans to release an improved version.

Google has released the open-source large model Gemna, breaking away from last year's closed-source strategy. Google is now engaging in a dual strategy of both closed and open source, competing in the large model market against OpenAI and Meta, forming a trio of giants: Google Gemna, Meta LLama, and Mistral. The competitive landscape in large models is intensifying as Google and Meta vie for dominance.

Google Gemini 1.5 Pro has raised concerns about the authenticity of videos generated by OpenAI's Sora, labeling them as forgeries and pointing out significant inconsistencies. The Gemini 1.5 Pro analyzes a scene featuring cherry blossoms in Japan, questioning the authenticity of the video. OpenAI's Sora is a text-to-video tool that generates highly detailed videos, garnering attention and being referred to as the ChatGPT moment in the field of video generation. Google Gemini 1.5 Pro boasts an impressive 1 million markers.

Google plans to launch Bard Advanced, a premium subscription service similar to ChatGPT Plus. Bard Advanced will be locked in subscription mode, providing more complex and responsive services. Gemini Ultra supports Bard Advanced and is expected to be released in early 2024. Bard has surpassed GPT-4, ranking second on third-party leaderboards with performance exceeding GPT-4. Google has canceled its contract with AI data company Appen, raising concerns about AI.