My AI Lawyer

The AI World Today

• Meta Launches Llama 3.1 Model

• Google’s Climate Forecasting Model

• Cohere’s Rerank 3 Nimble Model

• AI Competition Regulations Begin

• AI Disrupts Indian Economy, Survey

• UAE Launches Compass 2.0 Model

• AI Lawyer Launched in South Africa

  +

• Heads Up

• Save the Date

• AI Solutions

Meta Unveils Largest Open-Source Llama 3.1 AI Model

On July 23, Meta has released Llama 3.1, the largest open-source AI model to date, boasting 405 billion parameters. As per experts, this new model is the first time that a frontier-capability LLM is available to everyone to work with and build on. Llama 3.1 is seen to grade to GPT-4 and Claude 3.5 Sonnet, and the weights are open and permissively licensed, including commercial use, synthetic data generation, distillation and finetuning. Llama 3.1, trained with over 16,000 Nvidia H100 GPUs, is designed to cost roughly half as much to run in production compared to GPT-4o. Meta is collaborating with major tech companies like Microsoft, Amazon, and Google to help developers deploy custom versions of Llama 3.1. The model includes advanced features like integrating with search engine APIs and generating Python code for complex tasks. Available initially on WhatsApp and the Meta AI website, the assistant supports multiple languages and will soon expand to Instagram, Facebook, and Quest headsets. Despite its cost, Meta offers Llama 3.1 for free, aiming to make it the most widely used AI assistant by year-end.

Google Introduces Efficient NeuralGCM Weather Model

Google researchers have developed NeuralGCM, an AI-powered weather and climate model combining machine learning and neural networks with traditional physics-based modeling. NeuralGCM is more accurate than purely machine learning-based models for 1-10 day forecasts and surpasses top extended-range models. It is up to 100,000 times more efficient in simulating the atmosphere and can run on a laptop, unlike traditional models requiring supercomputers. NeuralGCM integrates a differentiable solver for atmospheric dynamics with machine-learning components, generating forecasts on par with leading methods. It accurately tracks climate metrics for decades and shows realistic tropical cyclone patterns. NeuralGCM offers significant computational savings and is open-source, making advanced weather and climate modeling more accessible.

Rerank 3 Nimble Boosts Enterprise Search Efficiency

Cohere, a startup that raised 500m dollars this week, has introduced Rerank 3 Nimble, the latest model in its Rerank series, offering a 3x increase in speed while maintaining high accuracy for enterprise search and Retrieval-Augmented Generation (RAG) systems. This new model, launched on July 23, supports over 100 languages and can handle very long documents. It excels in multi-aspect and semi-structured data reranking, making it ideal for high-volume workloads. Rerank 3 Nimble enhances search relevancy and reduces latency by a third compared to its predecessor. It's available on Amazon SageMaker and for on-premise deployments, with Amazon Jumpstart access starting July 30, 2024. Key applications include improving enterprise search systems, accurately searching complex documentation, and optimizing RAG systems.

US, EU, UK Issues AI Competition Joint Statement

Regulators in the United States, European Union, and Britain have signed a joint statement to ensure effective competition in the artificial intelligence sector, focusing on consumer protection. The statement emphasizes that generative AI's rapid evolution presents new competitive opportunities and technological inflection points, fostering innovation and growth. The involved watchdogs—the European Commission, the UK's Competition and Markets Authority, the U.S. Department of Justice, and the U.S. Federal Trade Commission—pledge to prevent tactics undermining fair competition. They aim to enforce principles of fair dealing, prevent exclusionary practices, and scrutinize investments and partnerships between established companies and newcomers. The regulators are committed to maintaining effective competition and ensuring fair treatment for consumers and businesses under their respective laws.

Economic Survey Highlights AI's Impact on Jobs in India

The Economic Survey 2023-24, presented in Indian Parliament, highlights AI as a major disruptor in India's economy. The survey notes that the Indian economy needs to create nearly 7.85 million non-farm jobs annually until 2030 to meet the rising workforce demands. AI, while posing risks, especially to the BPO sector, offers significant opportunities for productivity enhancement and job creation. The government has launched initiatives like 'Future Skills Prime' and 'YUVAi' to connect AI with youth. A budget of $1.25 billion for the India AI Mission aims to strengthen the AI ecosystem. The survey also emphasizes the need for an Inter-Agency Coordination Authority for AI to guide research and policy planning.

Core42 and Qualcomm Unveil Advanced Compass 2.0 Model

Core42, a subsidiary of G42 located in the UAE, and Qualcomm Technologies have officially launched Compass 2.0. Soft-launched in January, the platform has processed approximately 150 billion tokens. Compass 2.0, powered by Qualcomm Cloud AI 100 Ultra inference accelerators, sets new benchmarks in energy efficiency, flexibility, and cost-effectiveness for AI workloads. It integrates Qualcomm’s inference-as-a-service, supporting pre-trained models and scalable solutions within containerized environments. The platform expands API capabilities to include GenAI, embeddings, computer vision, and natural language processing models. Notable additions are JAIS, the leading Arabic large language model, and Azure OpenAI GPT-4 family models. This launch promises advancements in personalized conversational agents, patient interaction systems, recommendation engines, and text-to-image solutions.

Legal Interact Introduces AI-Powered Lawyer Service

Legal Interact, a South African IT service-based company, has launched an AI-powered online lawyer platform called My AI Lawyer. Utilizing Microsoft Azure and Azure Open AI, the platform offers legal intelligence to South Africans, aiming to enhance access to justice through technology. Co-founder Rapheal Segal highlighted the software's goal of providing better legal access, particularly for those without legal advice. The platform integrates with WhatsApp, allowing users to send questions and receive instant answers. It combines AI with human oversight to ensure compassionate, comprehensible, and well-thought-out responses. When necessary, it recommends human lawyers, balancing AI efficiency with human empathy.

Heads Up

AI EV: Chinese autonomous driving startup Pony.ai prepares for US IPO US IPO

AI Markets: Nvidia clears Samsung's HBM3 chips for use in China-market processor

AI Funding: Harvey raises 100m series C at 1.5 billion valuation

AI Poll: Musk launches poll asking if Tesla should invest $5 bln in xAI

AI Surveillance: Facial recognition and AI monitoring deployed in India’s largest railway network

AI Update: OpenAI assigns new project to AI safety leader Madry in revamp

AI Use: Only 20% of Australian organisations mature in AI adoption

AI 3D: Aibuild brings AI to large-format 3D printing in Australia

AI Education: UNESCO, Huawei to train East African university lecturers on AI

AI Invest: Foxconn to invest $138 mln for new business headquarters in China

AI Impact: 7 Top Investment Firms Using AI for Asset Management

Save the Date

October 10-11: Budapest Hosts European Rewards Conference 2024

AI Solutions

FaceSync: Animate Any Image with Your Performance

LensGo AI has introduced FaceSync, an innovative feature allowing users to record their performance videos and synchronize them with any image or video. By capturing facial expressions, lip movements, and voice, FaceSync maps these onto the chosen media, creating lifelike animations. This technology enables users to bring static images to life or animate videos while maintaining their voice and expressions. Ideal for content creation, education, entertainment, and marketing, FaceSync offers endless possibilities for dynamic and interactive experiences. Currently in testing, FaceSync promises to revolutionize digital media interaction, setting a new standard for personalized and immersive content.