Tencent Meeting AI Assistant optimizes meeting experience with natural commands, covering information extraction, content analysis, and intelligent reminders. It offers real-time meeting summaries and post-meeting follow-up organization. Built on Tencent's "Hybrid" language model, it enhances Chinese writing, logical reasoning, and task execution abilities for a streamlined meeting experience.

At Xiaomi 14 Ultra launch, the Imaging Brain received an upgrade, introducing "Xiaomi AISP" platform for AI model computational photography. Features include ULTRA RAW technology with 16-bit color depth, and ULTRA SNAP function for instant capture of 150 photos. Exciting breakthroughs in imaging technology.

Stability AI unveils Stable Diffusion 3.0, enhancing image quality, spelling, and performance in multi-topic prompts. New diffusion transformer architecture promises accelerated diffusion model performance. CEO notes improved spelling and coherency in text generation. Model parameters range from 800M to 8B, with plans for detailed technical reports. Upcoming video, 3D, and multimodal generation AI systems are also in the pipeline. Advanced granularity control features will be launched soon.

Microsoft has released the "Generative erase" feature, using AI to remove elements from photos in Windows Photos app. Available for Windows Insiders, with plans for Windows 11 and 10. No word on metadata or watermarks.

Nvidia's CEO highlights global AI tipping point with $22 billion revenue surge, surpassing $60 billion annual revenue with a 225% stock value increase. Challenges include supply chain issues and rising market competition.

Retell AI offers a revolutionary conversational voice API for creating natural-sounding voice agents. They focus on solving coordination issues in AI conversations, with an impressive 800ms response time. Features include voice stability control, real-time transcription, and upcoming emotion analysis. Developers can integrate their own frontend with Retell's platform, allowing for easy creation of voice agent prototypes through a no-code sandbox.

OpenAI's Sora, a new text-to-video model, integrates deep learning, NLP, and computer vision to generate detailed and coherent videos from text prompts. It surpasses previous technologies in handling visual data types, video length, and resolution. With a release date soon, Sora is currently accessible to experts, artists, and filmmakers.

Google has released "Help me write," a new AI feature for Chrome designed to assist users in writing or refining text based on webpage content. This tool offers writing suggestions and can understand the context of the webpage you're on. US users can enable it in Chrome settings and use it by right-clicking on text fields. Note: Do not provide personal information when using this tool.

Apple's $3,500 Vision Pro sparked reports of returns, but research suggests otherwise. Complaints included discomfort and lack of content. Apple aims to enhance user experience with future iterations.

The New York Times plans to test new ad targeting technology, GenAI. This move aligns with their belief in GenAI's potential impact on business. The tool recommends ad placements based on specific campaign goals, reaching niche audiences. The company is leveraging large language models to power this tool. They are recruiting advertising partners for closed beta testing in Q2. The tool aims to enable real-time campaign optimization.

Mistral AI unveiled innovative LLM prototype Mistral Next, excelling in logical reasoning, creative writing, and programming. It features ethical decision-making and plans for future updates, promising groundbreaking advancements in AI technology.

Microsoft is developing a new network card for the Maia AI chip to boost performance and reduce reliance on NVIDIA, optimizing the Azure cloud experience. Led by Pradeep Sindhu, the project aims to rival NVIDIA's ConnectX-7 card, with support from OpenAI and Azure Boost technology for significant gains in the AI software market. Acquisition of Fungible further strengthens Microsoft's position in data center infrastructure.

Google introduces new AI models Gemma 2B and 7B, surpassing Meta's Llama 2 and Mistral. Responsible AI Toolkit provided. Gemma can be used commercially with no restrictions. Support for popular frameworks and deployment on Google Cloud offered. Gemma enables diverse AI applications, with improved text generation and understanding capabilities. New possibilities for AI development highlighted by enhanced generation quality.

Match Group partners with OpenAI, buying over 1000 ChatGPT licenses. Employees will leverage AI technology to enhance products and user experiences, aiming for user growth by 2024. Trained employees can utilize ChatGPT-4 for tasks like coding and design, with a focus on responsible use.

Microsoft's CADO Mikhail Parakhin hinted at integrating OpenAI's Sora with Copilot, but it will take time. Sora is a chatbot creating detailed videos, while Copilot assists developers. Microsoft is still working on the technical aspects, with no specific timeline provided.

Google's Gemini 1.5 release introduces a 1M context window, sparking debates on Retrieval Augmented Generation (RAG). Gemini shows promise in handling complex text, but RAG's potential remains untapped. Gemini excels in retrieving key information, while RAG extends capabilities by accessing external sources. Notable figures like Oriol Vinyals acknowledge the benefits of both strategies, indicating a need for a balanced approach that leverages their respective strengths.

OpenAI's new AI model Sora impresses with lifelike videos. ChatGPT glitched on Feb 20, causing nonsensical outputs. Comparisons were made to "weird horror" graffiti. OpenAI quickly addressed the issue, but raised concerns about reliability for enterprise use.

Microsoft launched Copilot Pro, a $20/month subscription with AI features in Office apps, OpenAI models access, and improved image generation. Copilot Pro enhances Designer image tool, offers text and data generation in Word, PowerPoint, Excel, and Outlook, as well as GPTs for fitness, travel, and cooking. Copilot Pro shows promise in Microsoft's consumer AI efforts.

Apple is gearing up for its "AI year" in 2024, with iOS 18 set to introduce innovative artificial intelligence features like AppleGPT and Ajax framework. Speculations include AI-based Siri, Apple Music playlists, and more. CEO Tim Cook hints at further AI developments, though actual release timing remains uncertain.

Reddit signed a $60 million deal with an undisclosed AI company before its $5 billion IPO. This move could leverage user-generated content to train large-scale language models or develop new generative AI systems. However, Reddit's recent controversial decisions have faced backlash from the community. The latest Reddit data sale for AI training may intensify ethical concerns.