AI TOP Utility

The AI World Today

• Software Trains AI Locally Now

• Tencent’s Multilingual Singing Tech

• New AI Boost Medical Translation

• US, Canada Middle Market Embraces AI

• Do You Know About COPIED Act?

• Latin America's AI Regulations

  +

• Heads Up

• Save the Date

• AI Solutions

Gigabyte’s New Software to Train AI Models Locally

Taiwanese GIGABYTE TECHNOLOGY Co. Ltd has unveiled the AI TOP Utility, a revolutionary tool for local AI model training and fine-tuning, compatible with their new AI-targeted hardware. Designed for Linux, this utility supports models with up to 236 billion parameters, eliminating the need for remote data centers. Users can leverage GIGABYTE’s motherboards, AI TOP-branded SSDs, graphics cards, and power supplies for optimal performance. AI TOP simplifies AI model training with a user-friendly interface, customizable settings, and fine-tuning options. It supports over 70 open-source LLM models on Hugging Face, including Baichuan 2, Distill-GPT2, GLM4, Llama 2, and Llama 3. Unlike other AI software, AI TOP allows secure, local processing, ensuring data privacy. Additionally, the AI TOP Tutor provides setup guidance and technical support, making it accessible for both beginners and advanced users. This innovation democratizes AI development, making sophisticated AI model training more accessible and secure.

Tencent Unveils Multi-Lingual AI Singing Tech

Tencent Music-owned Kugou Music has unveiled KUGOU AIK, the world’s first multi-lingual AI singing technology. Users can sing a cappella for 60 seconds, and AIK can replicate their vocal characteristics to perform various songs. The technology supports 15 Chinese dialects and 10 languages, including English, Japanese, and Korean. On July 16, Irish pop group Westlife used KUGOU AIK to release a Mandarin cover of Jason Zhang’s 2016 hit, "Love + Courage," in China. Westlife, one of the most successful groups in the UK, supervised the creative process over several months. This innovative use of AI allows Western artists to engage fans in non-English markets, potentially transforming global music releases.

AI Amplified Introduces LLMs-in-the-loop for Healthcare

AI Amplified’s latest research introduces the "LLMs-in-the-loop" approach for medical translation, addressing unique global healthcare challenges. This methodology outperforms larger models like Google Translate and GPT-4-Turbo by creating specialized neural machine translation models. The key to their success lies in custom parallel corpora, compiled from scientific articles and medical texts in six languages, enhancing training accuracy. The innovative approach involves synthetic data generation, rigorous evaluation, and agent orchestration. This ensures precise and effective medical translations, crucial for disseminating knowledge and improving healthcare outcomes globally. The study highlights that small, specialized models can surpass larger, general-purpose ones in specific domains, making a significant impact on medical communication.

RSM Survey Highlights AI's Growing Impact In US, Canada

A new survey by AI consulting firm RSM highlights AI adoption trends among US and Canadian businesses. Released on July 18, the survey reveals that over 75% of middle-market companies are utilizing AI, with 77% adopting generative AI tools like ChatGPT and Microsoft Copilot. Despite high usage, more than half found implementation challenging, and 67% require external support for optimization. AI investments are on the rise, with 92% of US and 78% of Canadian companies planning budget increases. Key goals include improving quality control and automating tasks, underscoring AI’s growing impact on business operations.

New Bill In US Senate Targets AI Content and Provenance

On July 11, 2024, U.S. Senators Maria Cantwell, Marsha Blackburn, and Martin Heinrich introduced the Content Origin Protection and Integrity from Edited and Deepfaked Media Act (COPIED Act). The bill addresses the lack of transparency in AI systems and calls for consensus-based standards. It mandates the National Institute of Standards and Technology (NIST) to develop guidelines for detecting synthetic content and watermarking. The act also requires AI developers to offer content provenance options within two years and prohibits tampering with provenance information. The Federal Trade Commission (FTC) and State Attorneys General will enforce these provisions.

New Report Maps AI Regulation in Latin America

US-based Watchdog group, Access Now, has published the report "Regulatory Mapping on AI in Latin America," highlighting AI governance trends in the region. Released on July 18, the report reveals that while no Latin American country has a specific AI law, many have introduced regulatory proposals inspired by the EU Artificial Intelligence Act. Countries like Argentina and Brazil propose subject-specific regulations, while Chile, Costa Rica, and Colombia suggest establishing specialized authorities to oversee AI implementation. The report underscores the importance of aligning AI development with human rights frameworks, emphasizing that AI benefits should not compromise fundamental rights. This report is crucial for understanding the evolving AI regulatory landscape in Latin America and its alignment with global standards.

Heads Up

AI Funding: Dutch Chip Equipment Startup Nearfield Raises $148 million

AI Entertainment: ‘Big Brother 26’ Fans Have Already Identified AI Robot ‘Ainsle’ as Famous TikToker

AI Acquisition: Blend Acquires Nuvu, a Leader in Applied AI, in a Continued Latin America Expansion

AI Innovation: How a Saudi Startup is Using AI to Boost the Efficiency and Uptake of Solar Energy

AI Impact: Social media Conversations About AI Grow Over 386% as Indians Embrace New AI Era

AI Workspace: China's AI Giant iFLYTEK to Set Up International HQ in Hong Kong

AI Defense: US DoD AI Office Hosts First GIDE Industry Insights Forum

AI India: US-Headquartered AI Wealth Platform Backed by Top Asset Managers Launches in India

Save the Date

September 11-12: Don’t Miss Tech BBQ 2024 in Copenhagen

AI Solutions

JetLearn: Leading AI and Robotics Education for Kids

JetLearn, an Amsterdam-based online academy, has been recognized as a top EdTech startup in Europe by HolonIQ. Specializing in AI and robotics education for children and teens, JetLearn was honored in the “STEAM for Kids” category. The academy offers interactive courses in AI, coding, and robotics, designed to be fun and engaging. Using their AI platform, JetLearn provides live online lessons, workshops, and hackathon training, tailored to each student’s learning style. Founder Abhishek Bahl emphasizes preparing the next generation for an AI-driven future. HolonIQ’s recognition was based on product quality, team strength, and market momentum. JetLearn’s mission is to equip young learners with essential technology skills, impacting over 55 countries worldwide.