For Those Who Seek Unbiased News.

Be informed with 1440! Join 3.5 million readers who enjoy our daily, factual news updates. We compile insights from over 100 sources, offering a comprehensive look at politics, global events, business, and culture in just 5 minutes. Free from bias and political spin, get your news straight.

Join for free today!

Today:

• Google DeepMind Real-Time Videogame Creation

• Meta's Llama AI Dominates Market

• Apple, Nvidia Eye OpenAI Investment

• ChatGPT Hits 200M Weekly Users

• Nvidia’s Eagle AI Targets Jobs

• Oprah Interviews OpenAI’s Sam Altman

AI Doom | Google's STUNNING Videogame Generation Model BREAKS the Videogame Industry…

Nvidia CEO Jensen Huang predicted AI-generated games within 10 years, but Google DeepMind's new technology shows it's possible now. Using a neural model called "game engine," AI generates real-time gameplay without human coding. This innovation is showcased through Doom, a classic game known for its tech challenges. 

The AI learns from existing game footage, simulating the game as it plays. Reinforcement learning trains AI agents to gather data autonomously. The resulting gameplay is so realistic that human testers struggled to tell it apart from the original. 

WATCH THE VIDEO ON YOUTUBE

Meta leads open-source AI boom, Llama downloads surge 10x year-over-year

Meta’s open-source AI models, particularly the Llama family, are rapidly gaining traction, with downloads on Hugging Face surging tenfold to nearly 350 million in the past year. This growth, driven by the recent release of Llama 3.1, reflects a broader trend of open-source AI catching up to closed-source models in terms of performance and enterprise adoption. 

Major companies like Zoom, Spotify, and Goldman Sachs are now integrating Llama models into their operations. The rise of open-source AI is putting pressure on leaders like OpenAI to innovate and reduce costs, as open models offer greater flexibility and cost benefits.

VENTUREBEAT

Apple, Nvidia Are in Talks to Invest in OpenAI

Apple and Nvidia are in discussions to invest in OpenAI, joining a new multibillion-dollar funding round led by Thrive Capital that values the ChatGPT creator at over $100 billion. Microsoft is also expected to participate in the funding. 

This investment would deepen the tech giants' involvement with OpenAI, bolstering their competitive positions in the artificial intelligence race. The move signifies Apple and Nvidia's strategic alignment with OpenAI, as they seek to strengthen their AI capabilities alongside established partnerships.

THE WALL STREET JOURNAL

ChatGPT’s weekly users have doubled in less than a year

OpenAI’s ChatGPT now has over 200 million weekly users, doubling from 100 million last November. The surge in users highlights the growing popularity of the AI chatbot, despite competition from Google, Microsoft, and Meta. 

Additionally, 92% of Fortune 500 companies are using OpenAI’s products, and API usage has spiked with the release of the cost-effective GPT-4o Mini model. 

THE VERGE

Nvidia’s ‘Eagle’ AI sees the world in Ultra-HD, and it’s coming for your job

Nvidia has introduced "Eagle," a new AI model family that enhances machines' ability to understand visual information at ultra-high resolutions, up to 1024x1024 pixels. This multimodal model combines text and image processing, excelling in tasks like optical character recognition (OCR) and image segmentation. Eagle's high-resolution capabilities can significantly improve document processing in legal, finance, and healthcare sectors, reducing errors and costs. 

Released as open-source, Eagle aligns with Nvidia’s commitment to ethical AI, fostering transparency and collaboration. Its superior performance positions Nvidia as a key player in multimodal AI, potentially reshaping how AI interacts with the visual world.

VENTUREBEAT

Oprah Winfrey Sets Interview With OpenAI CEO Sam Altman for ABC Special on AI’s Impact

Oprah Winfrey will interview OpenAI CEO Sam Altman in a new ABC special, “AI and the Future of Us,” premiering on September 12. The special aims to explore AI’s impact on daily life, demystify the technology, and empower viewers to understand the AI landscape. 

The hour-long program will also feature Bill Gates discussing AI's revolutionary potential, FBI director Christopher Wray on AI’s use by criminals, and insights from YouTube creator Marques Brownlee and AI ethicists Tristan Harris and Aza Raskin. 

THE WRAP

🧠RESEARCH

Dolphin: Long Context as a New Modality for Energy-Efficient On-Device Language Models

The Dolphin model introduces a new way to handle long texts in language models on devices, reducing energy use and response time by 10x and 5x, respectively. By treating long text as a unique input type, it lowers the processing load, making AI more efficient for devices with limited resources.

Eagle: Exploring The Design Space for Multimodal LLMs with Mixture of Encoders

The Eagle study explores how to improve multimodal large language models (MLLMs) by using a combination of vision encoders to interpret complex visual information. Findings show that a simpler method—concatenating visual tokens—works as well as complex strategies. Eagle enhances model coherence and outperforms leading open-source MLLMs on key benchmarks.

The Mamba in the Llama: Distilling and Accelerating Hybrid Models

The paper presents a method to distill large Transformer models into linear RNN architectures like Mamba, reusing linear projection weights for efficient deployment. The hybrid model, retaining a quarter of the attention layers, matches Transformer performance in benchmarks and is faster with a new decoding algorithm. The approach enables efficient use of limited computational resources while maintaining high performance.

Text2SQL is Not Enough: Unifying AI and Databases with TAG

The paper introduces Table-Augmented Generation (TAG), a new method for answering natural language questions over databases, combining language models' reasoning abilities with data management systems' power. TAG addresses the limitations of current Text2SQL and Retrieval-Augmented Generation methods, enabling broader query types and opening new research areas. Standard methods currently answer only 20% of queries accurately, highlighting the need for further research.

LlamaDuo: LLMOps Pipeline for Seamless Migration from Service LLMs to Small-Scale Local LLMs

The LlamaDuo pipeline offers a solution for migrating from large cloud-based language models to smaller, local models, addressing privacy concerns, operational dependencies, and offline needs. By iteratively fine-tuning the smaller model with data generated by the service model, LlamaDuo ensures it matches or exceeds the original model's performance in specific tasks, providing a scalable AI deployment solution.

🛠️TOP TOOLS

Google AI Studio - Fastest way to start building with Gemini, our next generation family of multimodal generative AI models.

Haiper AI - Describe any video you can imagine with text.

CommandDash - Integrate any package, SDK or framework with expert AI agents. Get contexutalized code for your usecase within the IDE

AssemblyAI - Experience industry-leading speech-to-text accuracy with Speech AI models on the cutting-edge of AI research, accessible through a simple API.

Cursor - Built to make you extraordinarily productive, Cursor is the best way to code with AI.

📲SOCIAL MEDIA