TLDR AI 2024-07-22

Data breaches are in the news. But even if your password is safe, your personal data could be at risk. Data brokers are profiting from your home address, SSN, phone number, and other personal information - and they're selling it to scammers, spammers, and the highest bidder.

Incogni scrubs your personal info from the web. Simple as. It takes 3 minutes to set up, tackles 180+ data brokers and people-search sites, and there's a 30-day money back guarantee.

Protect your data, your time, and your sanity. Get Incogni today and never look back.

Use code TLDRAI today for a 60% discount →

Nvidia is developing a new AI chip, the B20, tailored to comply with U.S. export controls for the Chinese market, leveraging its partnership with distributor Inspur. Its advanced H20 chip has reportedly seen a rapid growth in sales in China, with projections of selling over 1 million units worth $12 billion this year. U.S. pressure on semiconductor exports continues, with possible further restrictions and control measures on AI model development.

The United Nations aims to create a centralized AI forum to address the fragmented global efforts in AI governance.

Former Google DeepMind scientists at Artificial Agency unveiled an AI behavior engine for dynamic NPCs in video games. They have raised $16 million to enhance in-game interactions. They're working with renowned AAA studios and expecting widespread adoption by 2025 despite potential cost implications. Their engine offers more realistic and responsive game characters without predefined scripts.

Larger models need a larger vocab, according to this new research which looks at scaling laws for vocabulary size.

Apple has released everything, including code and data, for its 7B parameter model, which is competitive with Mistral.

Researchers propose the Mixture of Multimodal Experts (MoME) to tackle the issue of task interference in generalist Multimodal Large Language Models (MLLMs).

OpenOCR is a unified training and evaluation benchmark for scene text detection and recognition algorithms.

Endia is an array based programming library for Mojo designed to assist in various scientific and machine learning tasks.

Txtai is an all-in-one embeddings database for semantic search, LLM orchestration, and language model workflows.

If you're looking to get in front of software developers, AI enthusiasts, executives and other tech professionals, consider advertising in TLDR. We have 10 newsletters to choose from and a subscriber base of 4+ million tech professionals to effectively reach your target audience. Learn more about running your first campaign with us.

Mantle utilized a Gemini 1.0 Pro LLM with a one million token window to convert a prototype project into a production-ready codebase, reducing the workload by handling boilerplate code and repetitive patterns. This approach, which leveraged extensive context and iterative code generation, saved months of developer time by allowing the team to focus on refining the crucial 20% of the project.

OpenAI's new gpt-4o mini model offers extremely low costs, enabling AI-generated content supported by ads to become financially viable. By generating dynamic blog posts for $0.00051525 each and earning approximately $0.0026 per ad impression, publishers can achieve a net profit of $0.002 per page view. This may cause a potential shift towards AI-generated content in response to user queries.

The new CerberusDet framework offers a flexible solution for object detection by combining multiple task heads within a single model based on the YOLO architecture.

This extremely simple CLI, which uses Claude Sonnet 3.5, can be used to suggest code edits to improve an existing code base.

People around the world want to learn new skills and master new technologies.

Taylor & Francis sold access to its authors' research to Microsoft for AI use without informing the authors or offering opt-out options, raising concerns over authors' rights.