UN Ethical AI, Microsoft AI Surface
March 22 2024: UN AI resolution, Microsoft's new AI PCs, Nvidia's AI games, AI deepfake legislation, AI agent outperforms nurses, Bengaluru AI traffic solutions
Good Morning & Happy Friday!
Latest in AI,
1 Minute Roundup
UN adopts the first formal AI ethics resolution.
AI models struggle with spelling and details, exposing comprehension limitations.
Microsoft launched the Surface Pro 10 and Laptop 6 with a 50% performance boost and AI features.
AI Deals: ValidMind ($8.1M), Sprih ($3M), Monaire ($3.5M), Rithmik Solutions (2M CAD), BorderlessAI ($27M), Zenapse ($8M), The Browser Company ($50M at $550M valuation), Profluent ($35M additional, total $44M), Presymptom Health (£1.5M), Clusterlab ($600K pre-seed), Quantmatix (€2.7M), Foundry ($80M), NLX ($15M), Tensorplex Labs ($3M), Cohere (in talks for $500M at $5 billion valuation).
Bipartisan bill for mandatory AI-generated deepfake labeling.
Bengaluru employs AI for traffic management and faces infrastructure challenges.
BBC explores AI for journalism and archive monetisation.
Nvidia CEO predicts AI-generated games within a decade.
LA schools deploy an "Ed" AI chatbot, which is a $6M investment.
Nvidia's AI healthcare agents cost $9/hour and outperform nurses.
Microsoft's $650M Inflection AI deal aims at tech licensing and investor returns.
Tennessee's ELVIS Act protects musicians from AI voice cloning.
Sakana AI launches open-source Japanese AI models.
GitHub's AI auto-fix tackles 90% of security issues.
Headlines
UN Adopts Pioneering Resolution to Guide Ethical AI Development
The UN General Assembly has unanimously passed a groundbreaking resolution promoting “safe, secure and trustworthy” artificial intelligence (AI) systems that will also benefit sustainable development for all.
Global Consensus: The resolution calls for AI technologies that respect online and offline human rights, mirroring rights protected in the physical world.
Developmental Equity: The Assembly encourages global collaboration to bridge the digital divide, ensuring equitable AI benefits for all, especially in developing countries.
Future Frameworks: The resolution sets a precedent for addressing AI's implications in other critical areas, including peace, security, and military applications.
Why AI Still Struggles with Spelling and Text Details
Spelling Woes: AI models like DALL-E and ChatGPT still struggle with accurately spelling words and generating coherent text details, often producing gibberish words like "taao" instead of "taco."
Pattern Matching: The underlying diffusion models for image generation and large language models (LLMs) for text do not natively understand rules like spelling or specific details. They reconstruct patterns from training data rather than truly "reading."
Language Complexity: While techniques like augmenting training datasets can help, resolving these issues comprehensively remains challenging due to the complexity of human languages.
Microsoft Launches AI-Powered Surface Pro 10 and Laptop 6 for Business
Microsoft has introduced the Surface Pro 10 and Surface Laptop 6, its first AI-driven PCs to transform the business landscape. These devices boast advanced Copilot AI features and significant performance upgrades.
AI Integration: The new Surface devices integrate Copilot AI, offering features like meeting summaries, customisable accessibility, and enhanced device security through AI-driven insights.
Surface Pro 10 Highlights: Powered by Intel Core Ultra processors and a neural processing unit (NPU), the Pro 10 delivers a 50% performance boost over its predecessor, with options for 5G connectivity, smart card reader and NFC and up to 64GB RAM. Starting at $1,199, it features a 140-degree HD camera and advanced inking technology.
AI Deals
ValidMind raised $8.1M in seed funding for its AI model risk management platform. The platform automates documentation and governance tasks in the financial services sector and streamlines the model validation process.
Sprih, an AI platform for sustainability, raised $3M in seed funding led by Leo Capital.
Monaire, an AI startup, raised $3.5M in a seed round led by Construct Capital to enhance HVAC/R maintenance in small commercial buildings.
Rithmik Solutions raised $2M (CAD) to expand globally and enhance its AI platform for reducing emissions in mobile mining equipment.
BorderlessAI secured $27M in funding, led by Susquehanna and Aglaé Ventures, to enhance its AI-powered HR platform.
Zenapse secured $8M in seed funding to enhance its AI SaaS marketing platform. The company aims to offer emotionally intelligent marketing experiences for brands like Sam’s Club and Aeropostale.
The Browser Company, maker of Arc browser, raised $50M in a funding round led by Pace Capital at a $550M valuation, aiming to redefine personal computing through the browser.
Profluent secured $35M to scale its AI-driven platform for biomedicine, focusing first on gene editing with CRISPR and advancing its mission of designing novel proteins for therapeutics.
Presymptom Health, an AI-based diagnostics startup for early detection of infections and sepsis, secured £1.5M.
Clusterlab secured $600K in pre-seed funding to advance Arabic AI tech. The company will focus on R&D for large language models to enhance products.
Quantmatix secured €2.7M in funding to accelerate the launch of its SaaS product for family offices and professional investors. The company leverages big data, AI, and machine learning for predictive analytics in trading.
Foundry secured $80M in Seed and Series A funding from Sequoia Capital and Lightspeed Venture Partners for its AI/ML public cloud.
NLX raised $15M in Series A funding to expand its AI-powered customer experience solutions for enterprise brands, including Red Bull and Copa Airlines.
Tensorplex Labs, a Web3 and AI startup, secured $3M in seed funding led by Canonical Crypto and Collab+Currency to develop decentralised AI infrastructure.
Cohere, an AI startup focusing on enterprise AI models, is in talks to raise $500M at a valuation of $5 billion.
Others
Bipartisan Bill Targets AI-Generated Deepfakes with Mandatory Labeling
A new bipartisan House bill aims to mandate the identification and labelling of AI-generated images, videos, and audio to counteract potential misinformation and exploitation risks associated with deepfakes.
The legislation proposes digital watermarks or metadata for AI-created content.
Tackling Bengaluru's Traffic Nightmare with AI
AI tools like congestion modelling, violation detection (10,000 daily), and 165 new adaptive traffic signals are used to tackle the Bengaluru traffic nightmare.
Despite AI's potential, experts warn that Bengaluru currently has only 200 cars per 1,000 people (“so-called” developed nations have 600-800). With no new roads for 24 years, AI alone can't solve the "vicious congestion circle”.
BBC's AI Ambitions: Developing In-House Models and Monetizing Archives
The BBC is considering partnering or going solo to build its AI models using decades of journalism content to power tools for in-house story assistance.
The broadcaster has held talks with tech giants like Amazon to monetise its archives by allowing access to training their "foundational" AI models.
Nvidia CEO Predicts AI-Generated Games Within the Next Decade
Jensen Huang, CEO of Nvidia, believes we could see fully AI-generated games in less than 10 years, potentially starting within the next 5 years.
Huang envisions future AI capabilities to generate games pixel-by-pixel, potentially leveraging cloud computing power. Interim steps may include AI-generated assets and NPCs before the entire game generation.
LA Schools Launch AI Student Advisor "Ed" for Academic Guidance
The Los Angeles Unified School District unveiled an AI chatbot called "Ed" to serve as a personalised student advisor, providing information on grades, attendance, assignments, readings, and non-academic matters like mental health resources.
Nvidia's $9/Hour AI Healthcare Agents Outperform Human Nurse
Nvidia and HippocraticAI have developed AI-powered "empathetic" healthcare agents that outperform human nurses in various tasks.
These AI agents cost only $9, compared to $39.05, the median hourly wage for human nurses (2022). It solves healthcare staffing shortages while enhancing patient access and outcomes.
Microsoft's Inflection AI Deal: A Strategic Move for Tech and Investors
Microsoft reportedly spent $650M on a deal with Inflection AI, securing non-exclusive tech licensing and safeguarding against legal disputes.
The deal aims to provide a positive return for Inflection AI's investors, with early backers receiving 1.5 times their investment and later investors getting 1.1 times.
Tennessee Enacts ELVIS Act to Safeguard Musicians Against AI Misuse
Tennessee's ELVIS Act, signed by Governor Bill Lee, protects musicians from AI misuse. It covers over 61,617 industry jobs and contributes $5.8 billion to the state's GDP.
The legislation updates personal rights to include voice protections against AI cloning.
Sakana AI Unveils Innovative Open-Source AI Models
Sakana AI, founded by former Google researchers in Tokyo, has released open-source Japanese language models. AI models have been developed through "model merging" and evolutionary techniques, leading to hundreds of model generations.
GitHub Enhances Security with AI-Powered Autofix Feature
GitHub's new "code scanning autofix" leverages AI to suggest fixes for over 90% of security issues. For languages such as JavaScript, TypeScript, Java, and Python, based on GitHub Copilot, CodeQL, and OpenAI GPT-4.
Love the AIML newsletter? Share it with your friends, family, and colleagues who want to stay on top of AI. Have a great weekend. Follow us.