SIGGRAPH 2024 💡, Meta SAM 2 🤖, AMD's Q2 2024 Earnings 📈

August 8th, 2024 GovAI Summit Reston, VA, United States Register Now

August 12-14, 2024 Ai4, MGM Grand, Las Vegas. Register Now

SIGGRAPH 2024: Mark Zuckerberg’s Vision for AI

At SIGGRAPH 2024, Mark Zuckerberg, founder of Meta, highlighted the growing integration of AI and VR in technology, underscoring Meta’s pioneering contributions through its AI Research Center, FAIR. In a conversation with Jensen Huang, CEO of Nvidia, Zuckerberg detailed Meta's advancements in computer vision, language models, real-time translation, and their development of PyTorch, a flexible platform for building neural networks.

Key Points:

1. Generative AI: Meta’s generative AI enhances content recommendation systems, improving user engagement on platforms like Instagram and Facebook by using generalized models.

2. AI Studio: Meta's AI Studio empowers content creators to build AI versions of themselves, enhancing interaction with their communities.

3. Open-Source Commitment: Meta’s open-source projects, such as PyTorch and LLaMA, have accelerated AI innovation and collaboration, benefiting the broader AI community.

4. Smart Glasses and AR: Meta is advancing AR technology through smart glasses, in collaboration with EssilorLuxottica, integrating AI capabilities for future holographic AR experiences.

5. Future Vision: Zuckerberg envisions a future with a unified AI model enhancing user experience and operational efficiency, integrating AI into daily life and business operations.

6. Personal Touch: The event also featured personal interactions between Zuckerberg and Huang, highlighting their camaraderie and shared excitement for the future of AI and technology.

Read the full version Here!

Meta Announces Segment Anything Model 2 (SAM 2)

Meta has introduced SAM 2, an advanced object segmentation model for images and videos. Key features include:

• Real-Time Segmentation: SAM 2 offers promptable segmentation, outperforming previous models in accuracy and speed.

• Zero-Shot Generalization: It can segment any object without custom adaptations.

• Open Source and Datasets: Available under the Apache 2.0 license, with the SA-V dataset under CC BY 4.0, and a web demo provided.

• Impact: Revolutionizes object segmentation, inspiring AI advancements in various fields.

• Technical Highlights: Features promptable segmentation, mask decoder, and real-time processing. Trained on extensive datasets for superior performance.

• Future Work: Plans to enhance handling of challenging scenarios like occlusions and crowded scenes.

Meta aims to foster innovation by sharing SAM 2 and its resources with the AI community.

Read the full version Here!

AMD is becoming an AI chip company, just like Nvidia?

AMD's Q2 2024 earnings revealed that nearly half of its sales now come from data center products, marking a significant shift from its traditional focus on PCs and gaming consoles. The AMD Instinct MI300 accelerator drove this growth, generating over $1 billion in sales this quarter, doubling the company’s data center business over the past year. This performance aligns AMD more closely with Nvidia, which has similarly capitalized on the AI chip market.

To compete with Nvidia, AMD plans to release new AI chips annually, including the MI325X in Q4 2024, the MI350 in 2025, and the MI400 in 2026. CEO Lisa Su indicated that the MI350 will be competitive with Nvidia's Blackwell chip. Despite supply chain improvements, AMD expects supply to remain tight through 2025 due to high demand.

Overall, AMD's revenue for the quarter reached $5.84 billion, with a 19% increase in net profit. The company's personal computing segments also showed positive performance, with Ryzen CPU sales up 49% year-over-year and Radeon 6000 GPU sales growing. Upcoming Zen 5 laptops featuring Ryzen AI 300 "Strix Point" chips from brands like Acer, Lenovo, Asus, HP, and MSI are expected to bolster this growth further. This strong performance has positively impacted investor sentiment, positioning AMD as a rising competitor in the AI chip market.

OpenAI launches experimental GPT-4o Long Output model with 16X token capacity
Yesterday, the company discreetly unveiled a webpage introducing a new large language model (LLM): GPT-4o Long Output. This model is a variation of the GPT-4o released in May, but with a significantly increased output capacity—up to 64,000 tokens compared to the original GPT-4o’s 4,000 tokens, marking a 16-fold increase.

China stops worrying about lack of GPUs and learns to love the supercomputer
China is reportedly developing a new supercomputer that does not rely on GPUs, typically supplied by Western companies like NVIDIA, which have been subject to export restrictions. Instead, China is focusing on creating its own chip technology for this project. This move aligns with China’s broader goal of achieving technological self-sufficiency, especially in high-performance computing and AI, where Western sanctions have impacted its access to cutting-edge technology. The supercomputer aims to advance China's capabilities in AI and scientific research.


How to stop Elon Musk’s Grok from training on your X posts
To stop Elon Musk's AI platform, Grok, from using your posts on X (formerly Twitter) for training, you can make your account private. This prevents the platform from accessing your content for AI model training. If your account is public, your posts may be used unless X introduces specific settings to opt out. Users concerned about privacy or data usage should consider adjusting their account settings accordingly.


OpenAI has released a new ChatGPT bot that you can talk to
OpenAI has introduced a new version of ChatGPT that allows users to interact with the bot through voice conversations. This feature is part of the broader effort to make AI interactions more natural and accessible, moving beyond text-based communication. The new bot uses advanced voice synthesis technology, enabling users to have fluid and dynamic conversations with the AI. This update is seen as a significant step toward integrating conversational AI into everyday life.


Meta Discontinues Celebrity AI Chatbots
Meta has decided to discontinue its celebrity AI chatbots, which were introduced less than a year ago. These chatbots, featuring AI personas based on celebrities, were part of Meta's attempt to engage users on its platforms. However, the company is shifting its focus to other AI projects and features, leading to the end of this particular initiative.

👥 Connect & Feedback!

👉 Join Us:

📧 Advertise In Weekly AI News:

📧 Contact directly at [email protected]

😍 Share with your friends!