AI News The Gemini 1.5 pro has been in private preview in Vertex AI 1.5 Pro shows dramatically enhanced performance and includes a breakthrough in long-term context understanding, which means it can run 1 million tokens of information consistently, opening up new possibilities for enterprises to create, discover, and build using AI .
Gemini has multimodal capabilities and can process audio video text code .For example, Gaming company could now offer video analysis to support players, including tips to improve performance or an insurance company could combine video image and text to create an incident report and automate the claims process .SOURCE
“AI Hypercom and Nvidia Lead the Charge in Advanced AI Hardware Innovations”
AI News AI Hypercom orchestrates an impressive array of hardware and software, from programming languages and compilers to the chips and networks that drive them. This system-level integration offers efficiency gains up to twice that of baseline solutions that focus solely on hardware delivery. Our strategic approach, combining visionary planning with precise execution, has earned widespread recognition.
Nvidia continues to expand its GPU portfolio, announcing the general availability of the A3 Mega, powered by Nvidia’s H100 Tensor Core GPUs. These new units double the network bandwidth per GPU compared to previous A3 instances, marking a significant leap in performance.
Looking ahead, Nvidia’s latest Grace Blackwell series of GPUs is set to arrive at Google in early 2025. This lineup includes the Nvidia B200 and GB200 chips, powered by next-generation technology. The GB200 stands out as one of the most advanced chips globally, requiring liquid cooling for peak efficiency due to its high performance capabilities.
Moreover, our general TPU pool now consists of 8,960 interconnected chips, supporting large-scale machine learning training and applications. The TPU v5p pods have quadrupled the compute capacity per pod compared to our earlier models, enhanced further by TPU multislice technology, setting new standards for computational power in machine learning. SOURCE
Google Axian: Revolutionizing Data Center Efficiency and Speed”
AI News We’ve upgraded our TPUs with Gemini models to boost AI training and inference speeds, and we’ve enhanced our storage solutions. Today, we’re excited to introduce Hyperdisk ML, which accelerates inference significantly.
Additionally, we’re unveiling the Google Axian processor, our first custom ARM-based CPU designed specifically for data centres. Set for preview later this year, the Google Axian merges Google’s AI know-how with the latest compute core designs, delivering performance up to 50% better and energy efficiency up to 60% greater than current generation x86-based VMs. SOURCE
“Vertex AI: Tailoring Advanced AI Models to Fit Every Enterprise Need”
AI News Vertex AI, recognized as the fastest-growing enterprise AI platform, offers access to over 130 models, including the latest Gemini partner models like Clude from Anthropic and popular open models such as Llama, Gemma, and MOL. These are available in various configurations, allowing you to select the ideal model tailored to your specific use case, budget, and performance requirements. With Vertex AI, you can seamlessly switch between models as your needs evolve. SOURCE
“Unveiling Gemini 1.5 Pro: Advanced AI for Optimal Data and Workplace Efficiency”
AI News Gemini 1.5 Pro now features the world’s largest context window, supporting up to 1 million tokens. This allows customers to process vast quantities of data in a single stream, including an hour of video, 11 hours of audio, codebases with over 30,000 lines of code, and more than 700,000 words. Its enhanced capabilities enable comprehensive cross-modality analysis.
Furthermore, we’re introducing Gemini Employee Agents for Google Workspace and Vertex AI, designed to boost productivity and collaboration. Embedded directly into Gmail, Docs, Sheets, and more, these AI-powered agents are revolutionising workplace efficiency. Today, we’re excited to announce several updates to Gemini for Workspace. Responding to customer feedback, we’re now offering AI-powered Google Meet and Chat at a more affordable price, providing a compelling alternative to platforms like Zoom. In a recent benchmark study, Google Meet surpassed competitors like Microsoft Teams, Zoom, and WebEx in video and audio performance.SOURCE
“New AI Add-ons Boost Workspace Security and Efficiency”
AI News Introducing the new AI Meeting and Messaging Add-on, designed to enhance productivity with capabilities such as automated note-taking, efficient chat summarization, and seamless real-time translation. This add-on is available for just $10 per user per month, offering an affordable solution to improve communication across global teams.
We are also excited to announce the AI Security Add-on for Workspace. This innovative feature empowers administrators to automatically classify and protect sensitive files and data. It utilizes a privacy-preserving AI model, custom-trained for each organization, to ensure that your critical information is secured according to your specific compliance and privacy requirements. SOURCE
“Quick and Secure Proposal Reviews with Gemini Workspace”
AI News Gemini Workspace is streamlining customer workflows by saving significant time on evaluating proposals. Without needing any preparatory work, users can access vendor proposals directly in Google Drive, assured by top-tier security protections. Simply drag and drop the proposal files into the Gemini side panel—currently in preview—and within seconds, Gemini will generate a concise outline of the proposals, providing a clear understanding of their content. This tool can even compare pricing details across documents. What used to take hours to review can now be summarized swiftly by Gemini, offering quick and accurate insights directly in the Gemini side pane. SOURCE
Adobe purchases videos at $3 per minute to drive AI
AI News Adobe Inc. is ramping up its investment in video content, spending $3 per minute to enhance its artificial intelligence (AI) text-to-video generator, as it seeks to compete with OpenAI’s Sora. The company is leveraging its vast network of photographers and artists, encouraging them to submit video clips that depict common activities and a range of human emotions like walking, joy, and anger. These efforts are part of Adobe’s strategy to build a rich library of assets to train its AI systems, according to the company.
Over the past year, Adobe has significantly expanded its creative software suite, integrating generative AI capabilities into flagship products such as Photoshop and Illustrator. These tools, which can convert text into images and illustrations, have achieved billions of engagements, as reported by Bloomberg.
Despite these advancements, the launch of OpenAI’s video-generation tool, Sora, has sparked concerns among Adobe’s investors about the company’s ability to keep pace with such innovative disruptions. In response, Adobe has reaffirmed its dedication to advancing its video-generation technology, with plans to unveil more developments later this year.
For its latest project, Adobe is requesting over 100 short video clips that showcase basic human actions, emotions, and simple anatomical details, as well as interactions with everyday objects like smartphones and fitness equipment. Contributors are advised to avoid copyrighted or offensive materials.
Participants in this project can earn an average of $2.62 per minute for their footage, with the potential to make up to $7.25 per minute.
This initiative underscores the extensive data requirements essential for powering popular AI-driven content creation tools such as ChatGPT. The ongoing debates over data sourcing have been highlighted recently, with Mira Murati, OpenAI’s CTO, discussing Sora’s training data sources in a widely noted interview with The Wall Street Journal, suggesting platforms such as YouTube, Facebook, and Instagram might be involved.
Adobe, however, is taking a different approach by primarily using its vast stock media library, which is specifically tailored to meet the needs of marketers and creative agencies, to train its AI models. SOURCE
Conclusion
This compilation of groundbreaking developments highlights a transformative period in artificial intelligence and technology. From Adobe’s strategic enhancement of its AI video capabilities to Google’s launch of advanced Gemini 1.5 Pro models and new AI-driven workplace solutions, the pace of innovation is accelerating. Each of these initiatives showcases the industry’s commitment to pushing the boundaries of what AI can achieve, from improving personal productivity to enhancing enterprise efficiency and securing sensitive data.
Furthermore, Nvidia’s advances in GPU technology and Meta’s anticipated launch of Llama 3 reflect the ongoing competition and innovation within the hardware and generative AI spaces, respectively. These efforts are not just about advancing technology but also about making these powerful tools more accessible and effective for a broader range of applications and users.
As AI continues to evolve, the integration of these technologies into everyday applications and systems underscores the critical importance of balancing innovation with ethical considerations and privacy concerns. The future of AI is being shaped by these developments, promising not only smarter tools but also a new paradigm for human-computer interaction and digital creativity.
Stay connected with Arcot Group for more AI News into how such collaborations are reshaping the tech landscape and paving the way for future innovations. For further reading on similar breakthroughs and the impact of AI and robotics, explore our blog.
