Musk plans xAI supercomputer, Microsoft reveals AI-powered PCs, Ray-Ban | Meta Smart Glasses Add Amazon Music, Google's AI Overview issues....

Microsoft reveals AI-powered PCs

Microsoft has unveiled a new category of AI-powered PCs branded as "Copilot+PC," marking a significant step in the integration of artificial intelligence into personal computing. These devices leverage both on-device AI capabilities and cloud-based large language models to enhance performance and user experience.

• Recall, a Windows 11 tool that remembers user activity to make it searchable.
• GitHub Copilot Extensions, for third-party software to add new skills to GitHub Copilot.
• Team Copilot, to work across Microsoft Teams.
• Custom?Copilot AI agents?for developers and businesses.
• MS Paint?is getting an AI image generator.
• Azure AI Studio, now with support for GPT-4o and a multimodal Phi-3.
• Real-time?translation and dubbing?for videos in Microsoft Edge.

• Pre-orders began on May 20, 2024 with prices starting at $999. Devices will be available from June 18, 2024.
• Microsoft has partnered with Acer, ASUS, Dell, HP, Lenovo, and Samsung to produce Copilot+PCs, in addition to its own Surface devices.

Nvidia Smashes Records, Stock Soars Past $1,000 on Surging AI Demand

Nvidia reported stellar Q1 2024 results, with revenue of $26.04 billion, ?driven by surging data center revenue of $22.56 billion?

Key highlights:

1. Strong demand for AI applications fueled data center growth, particularly Hopper GPU shipments.
2. Announced 10-for-1 stock split effective June 7 and 150% dividend hike to $0.10/share.
3. Expects substantial Blackwell AI GPU revenue in 2024, driving further growth.

Google's AI Overview Issues

Google's new AI Overview feature, which generates summaries at the top of search results, has come under fire for producing inaccurate, misleading, and even dangerous information. The AI-powered tool has made numerous factual errors and cited questionable sources like satirical articles and Reddit jokes, leading to widespread criticism and concern about the reliability of Google's AI systems.

Google's AI Overview feature has generated concerning instances of misinformation, including:

• Suggesting adding non-toxic glue to pizza sauce, based on an 11-year-old Reddit joke
• Advising that it's safe to leave dogs in hot cars
• Recommending staring at the sun for 5-30 minutes for health benefits, falsely attributing this to WebMD
• Proposing eating rocks daily, citing a satirical article from The Onion as if it were factual

These examples highlight the risks of AI systems confidently presenting false information as truth, especially when citing authoritative sources.?The spread of such AI-generated misinformation, whether intentional or due to AI "hallucinations," poses a growing threat as these tools become more prevalent.

Meta and Google want to make AI deals with Hollywood studios

Meta and Google are offering Hollywood studios substantial sums to license content for improving AI-generated video models. However, studios have shown mixed interest, with concerns over AI use of their creative work.

Key Points:

• Meta and Google are offering Hollywood studios tens of millions of dollars for licensing deals to enhance AI video models.
• There is an AI industry trend to secure media content, highlighted by OpenAI's multi-year deal with NewsCorp and Meta's consideration of paying publishers for content.
• Hollywood studios have reservations about AI companies using their creative work, as shown by Scarlett Johansson's accusation against OpenAI for allegedly copying her voice without consent.

Elon Musk plans xAI supercomputer

Elon Musk is planning to develop a supercomputer for his artificial intelligence company, X.AI, which aims to compete with leading AI firms like Google, Microsoft, and OpenAI. This supercomputer, potentially called Dojo2, is part of Tesla's broader AI ambitions and is expected to focus on large models, leveraging Tesla's extensive driving data for training.

Tesla plans to invest over $1 billion in the Dojo supercomputer, which will use in-house hardware and NVIDIA GPUs to achieve a target of 100 exaflops of AI computing power by next year. X.AI is also in talks to raise up to $6 billion from investors, targeting family offices in Hong Kong and sovereign wealth funds in the Middle East, aiming for a valuation of $20 billion.

This Week's Essential Read

Managing extreme AI risks amid rapid progress : New AI safety paper in Science

The article "Managing Extreme AI Risks Amid Rapid Progress," published in Science, addresses the significant risks associated with the rapid advancement of AI technologies and proposes measures to manage these risks effectively. Here are the key points:

1. Rapid AI Progress and High Stakes: AI is advancing quickly, potentially surpassing human abilities in various tasks. While this progress offers substantial benefits, it also presents significant risks, such as social harm, loss of human control over autonomous systems, and amplification of social injustices
2. Reorienting Technical R&D: Current AI systems lack essential safety capabilities. More focused research is needed on AI oversight, robustness, interpretability, and ethical use to ensure these systems do not pursue unintended harmful goals
3. Governance Measures: Effective governance frameworks are crucial. National and international institutions must enforce standards to prevent misuse and ensure AI advancements are safe and ethical. Policies should adapt automatically as AI capabilities increase, requiring proactive risk management
4. Risks of Autonomous AI: Autonomous AI systems could act independently in ways that are difficult to control, potentially leading to large-scale social manipulation, automated warfare, and surveillance. There is a pressing need to understand and mitigate these risks before they become unmanageable
5. Urgency and Preparedness: The scale of AI risks necessitates immediate and proactive measures. Investing in AI safety and ethical research is critical to steer AI development towards positive outcomes and avoid catastrophic scenarios