Smart Cop

The AI World Today

• NVIDIA Embraces Open Source

• Hugging Face’s AI Model For Mobile

• Meet Microsoft Designer App

• Trump's AI Manhatan Projects

• Meta's AI Regulatory Standoff

• Google’s New India Initiatives

• Meet Abu Dhabi's Smart Cop

• AI Transforms African Economy

  +

• AI Spotlight

• Heads Up

• Save the Date

• AI Solutions

NVIDIA Releases Open-Source GPU Kernel Modules

On July 17, NVIDIA announced the full transition to open-source GPU kernel modules with its upcoming R560 driver release. This change signifies a major step towards enhancing the integration and performance of NVIDIA GPUs on Linux systems. Over the past two years, NVIDIA has worked on developing these open-source modules, initially targeting data center GPUs with GeForce and Workstation GPUs in alpha state. The R515 driver was a milestone in this effort. Key improvements include heterogeneous memory management, confidential computing support, and enhanced performance, making the open-source modules competitive with proprietary versions. This move is expected to boost developer capabilities in debugging and integration, improve user experiences, and ensure tighter OS integration. The transition also holds significant implications for the tech industry, especially for enterprise and data center use. It addresses security concerns related to closed-source drivers and enhances support for confidential computing. Moreover, it aligns NVIDIA with industry trends towards open-source software, potentially influencing other companies to adopt similar strategies.

Hugging Face Releases Superior Compact AI Models

Hugging Face has unveiled SmolLM, a new family of compact AI language models designed for personal devices, outperforming offerings from Microsoft, Meta, and Alibaba’s Qwen. SmolLM includes three sizes—135 million, 360 million, and 1.7 billion parameters—delivering superior performance in common sense reasoning and world knowledge benchmarks. The models owe their success to high-quality data curation and transparency in the development process. SmolLM can run on personal devices, ensuring privacy and reducing costs. This release democratizes AI, making it accessible and privacy-friendly, and allowing developers to create custom AI applications without expensive cloud infrastructure.

AI-Powered Microsoft Designer App Goes Live

Microsoft’s AI-powered Designer app has officially launched for iOS, Android, and Windows. Previously in preview, Designer offers templates for creating custom images, stickers, greeting cards, and more. It leverages AI to edit images, restyle them, and create collages. Users can now access it through a personal Microsoft account. Key features include AI-driven image editing, background removal, and an avatar creator. Integrated into Copilot for Word and PowerPoint, it enhances document design capabilities. Designer provides 15 daily boosts for free, with an option to upgrade to Copilot Pro for 100 daily boosts.

Trump Allies Plan Sweeping AI Executive Orders

Former President Donald Trump’s allies are drafting an extensive AI executive order aimed at launching "Manhattan Projects" for military technology and reducing regulations. This plan, revealed by The Washington Post, includes establishing "industry-led" agencies to evaluate AI models and protect systems from foreign threats. The initiative, supported by the America First Policy Institute, signals a shift towards AI policies favorable to Silicon Valley. It also plans to repeal Biden's AI executive order, aiming to boost AI development and align with national security and economic goals. This marks a significant potential change in U.S. AI strategy.

Meta Halts AI Tools in Eropean Union and Brazil

Meta will withhold its upcoming multimodal AI model and future ones from the European Union due to regulatory uncertainties, Axios reports. This decision follows Meta's frustration with the lack of clarity from EU regulators, contrasting with the UK where similar laws exist but with clearer guidance. Meta's new model, which integrates video, audio, images, and text, will be available globally but not in the EU. Coupled with that, Meta has also suspended generative AI tools in Brazil following government objections. This move highlights a growing tension between Meta and international regulators. Brazil, a key market for Meta with over 200 million people, saw its National Data Protection Authority (ANPD) suspend Meta's new privacy policy related to AI training. This suspension impacts Meta's AI-driven ad targeting on WhatsApp, launched in June. Meta argues that training on regional data is crucial for product relevance, while regulators emphasize data protection and privacy compliance.

Boosting Indian Startups with Google's AI Program

Google, in partnership with MeitY Startup Hub, has announced an initiative to train 10,000 Indian startups in AI. This effort aims to democratize AI and support India's startup ecosystem. The initiative includes providing up to $350,000 in Google Cloud credits to eligible startups, AI-first programming, and curriculum through Startup School and Appscale Academy. Google is also launching AI innovation programs, such as a nationwide Gen AI Hackathon and the Solve for India Startup Bootcamp. Additionally, Google is expanding access to its AI models and optimizing AI for mobile devices to enhance on-device capabilities.

AI-Powered UAE Traffic Robot Enhances Road Safety

The Abu Dhabi Police in the United Arab Emirates (UAE) have introduced an AI-powered traffic police robot to enhance road safety. This human-like robot, part of the Directorate of Traffic and Security Patrols' initiative, aims to educate and engage the public on traffic safety. Programmed by the country’s national staff, it displays digital traffic awareness videos, responds to public inquiries, and offers traffic tips. The robot can also participate in educational lectures and traffic safety contests at events. This deployment reflects the leadership’s vision for a sustainable future and aligns services with global digital traffic education trends.

AI Boosts Africa's Economy $2.9 Trillion by 2030

The GSMA report, “AI for Africa: Use cases delivering impact,” published in July 2024, highlights that AI could boost Africa’s economy by $2.9 trillion by 2030. GSMA, a global association representing mobile operators, collaborated with AI4D Africa on this study. It identifies over 90 AI use cases in Kenya, Nigeria, and South Africa, particularly in agriculture (49%), climate action (26%), and energy (24%). AI-driven solutions, such as machine learning advisory services, are enhancing productivity and addressing energy access challenges. Future prospects include mobile-based edge computing, expected to drive AI proliferation as smartphone penetration rises to 88% by 2030, fostering economic growth across sectors like agriculture, healthcare, and climate resilience.

AI Spotlight

Assembly AI: Create a Multilingual AI Voice Translator

To create an AI voice translator that keeps your voice in different languages, follow this guide. The guide uses AssemblyAI for transcription, a translation module in Python, and ElevenLabs for text-to-speech synthesis. Below is a concise training guide:

Setup and Dependencies

1. Install Libraries: Gardio Assemblyai, Eleven Labs, and Google Trans

2. Get API Keys:

   • Sign up for AssemblyAI and ElevenLabs to obtain API keys.

Building the App

1. Gradio Interface:

   • Create a simple Gradio interface for audio input and outputs.

2. Audio Transcription:

   • Use AssemblyAI to transcribe the audio.

3. Text Translation:

   • Translate the transcribed text using Googletrans.

4. Text-to-Speech Synthesis:

   • Use ElevenLabs to synthesize speech in your voice.

Putting It All Together

• Integrate transcription, translation, and synthesis into the voice_to_voice function.

Running the App

• Launch the Gradio interface.

This guide provides the basic steps to build an AI voice translator using AssemblyAI, Googletrans, and ElevenLabs. For detailed code and examples, refer to the official documentation and the tutorial video on YouTube.

Heads Up

AI Regulation: US Legislation To Curb AI Deepfakes, Give Control Back To Creators

AI Competition : US Considers Tougher Trade Rules in China Chip Crackdown

AI Invest: Saudi Aramco’s Bold Footprint In Silicon Villy, Backing AI startup

AI Innovation:HP’s New AI Computer Raises the Stakes in the Battle of Tech Hardware

AI Capital: Introducing Anthology Fund: $100M Partnership Between Menlo Ventures, Anthropic

AI Funding: UAE-based AI Start-up Sentient Labs Raises $85mln in Seed Funding

AI Service: Samsung to Launch AI-Powered Air Conditioners in North America

AI Africa: Kenya Beats Nigeria in AI investment Inflows

AI Education: E&, Samsung Partner to Launch Smart Schools Initiative in UAE

In Funding: CytoReason Secures $80M to Scale AI Models, Establish Cambridge Hub

AI Robotics: Indian Firm Launches AI-Powered Robot For Global Market

AI Acquisition: Samsung Agrees to Acquire British Startup Oxford Semantic for AI

Save the Date

July 31: Deadline for Fatima Fellowship 2024 Application

AI Solutions

Meet OpenAI’s Affordable GPT-4o Mini Model

OpenAI has unveiled GPT-4o mini, a cost-efficient, smaller AI model and tool designed to make advanced AI technology more affordable and energy-efficient. Priced at 15 cents per million input tokens and 60 cents per million output tokens, GPT-4o mini is over 60% cheaper than GPT-3.5 Turbo. It outperforms the GPT-4 model on chat preferences and scores 82% on the Massive Multitask Language Understanding (MMLU) benchmark, surpassing Google's Gemini Flash and Anthropic's Claude Haiku.

The GPT-4o mini supports text and vision in its API, with future updates planned to include text, image, video, and audio inputs and outputs. ChatGPT's Free, Plus, and Team users can access GPT-4o mini starting Thursday, replacing GPT-3.5 Turbo, while enterprise users will gain access next week. This model targets a broader customer base, including companies with limited resources looking to deploy generative AI in their operations.