Meet Apple Intelligence

Sign Up | Advertise

The AI World Today

• Apple Intelligence Now Available

• Meta’s New AI Search Engine

• 200,000 GPUs Power xAI's Colossus

  +

• Heads Up

• AI Solution

Apple Intelligence Launches with iOS 18.1 Update

Apple has released iOS 18.1, iPadOS 18.1, and macOS Sequoia 15.1, introducing its anticipated Apple Intelligence features. These generative AI capabilities, first previewed at WWDC in June, are now available on select devices, including the iPhone 15 Pro/Pro Max, all iPhone 16 models, iPads with A17 Pro or M1 chips, and Macs with M1 or later. After updating, users can opt in to a waitlist to access these features, which rely on both local processing and Apple’s remote servers. Initial tools include writing assistance, image cleanup, article summaries, and a revamped Siri typing input. Additional AI features like Genmoji, Visual Intelligence, and ChatGPT integration are set for the upcoming iOS 18.2 update, now in developer beta.

Meta Develops AI Search, Reduces Google Reliance

Meta is reportedly developing its own AI-powered search engine to reduce reliance on Google and Microsoft. Currently, Meta’s AI chatbot relies on Google and Bing for real-time information on news, sports, and stocks. The new search engine, designed to provide AI-generated summaries of current events, would integrate seamlessly into Meta's platforms, including Instagram and Facebook. Meta’s efforts include web crawlers and a robust database, supporting the AI chatbot to deliver real-time answers without external dependencies. Additionally, Meta is gathering location data, hinting at future competition with Google Maps. This shift aligns with industry moves—such as Apple’s similar efforts and OpenAI’s upcoming SearchGPT—to offer proprietary search solutions, streamlining in-house AI ecosystems across companies.

xAI's Colossus Supercomputer Doubles to 200,000 GPUs

xAI is expanding its Colossus supercomputer to 200,000 NVIDIA Hopper GPUs, doubling its capacity as the world’s largest AI supercomputer. Located in Memphis, Tennessee, Colossus is built in collaboration with NVIDIA and leverages the NVIDIA Spectrum-X™ Ethernet networking platform, achieving 95% data throughput with zero packet loss. This powerful infrastructure enables xAI to train its Grok family of large language models, accessible to X Premium subscribers. Impressively, the facility was completed in just 122 days, with training commencing within 19 days of setup. The network's Spectrum SN5600 Ethernet switches and BlueField-3® SuperNICs deliver exceptional low-latency performance, supporting advanced AI workloads. Colossus exemplifies scalable, high-performance networking critical for multi-tenant generative AI systems.

Heads Up 

Google Search’s AI Overviews, providing quick snapshots of information atop search results, are now available in over 100 countries, the company announced Monday.

Universal Music Group and Klay Vision are developing KLayMM, an “ethical” AI model for music generation, aiming to launch within months to revolutionize music creation.

Alibaba released FunASR-1.1.13, boosting VAD model inference speed by 30%, and introduced funasr-runtime-sdk-online-cpu-0.1.12 for real-time transcription in multiple languages.

Sierra, an AI startup co-founded by OpenAI chairman Bret Taylor, raised $175 million in funding, reaching a $4.5 billion valuation.

Toyota and NTT will invest 500 billion yen ($3.26 billion) in AI software R&D for self-driving advancements, according to a Nikkei report Tuesday.

OpenAI CFO Sarah Friar announced that 75% of revenue originates from paid ChatGPT subscriptions, with a 5–6% conversion rate from 250M weekly active users.

AI Solution

Google’s InkSight Converts Handwriting Photos to Digital Ink

Google's InkSight, an innovative model, converts photos of handwriting into “digital ink” without requiring specialized equipment. Designed to bridge the gap between traditional handwriting and digital note-taking, InkSight captures stroke-level pen trajectories to create editable digital representations of handwritten text. Unlike optical character recognition (OCR), which transcribes text, InkSight recreates the strokes, preserving users' handwriting style for more realistic digital notes. The model uses a multi-task training setup, combining recognition and derendering, allowing it to work across various image conditions. It includes advanced techniques like OCR-based bounding box extraction and data augmentation for improved accuracy. InkSight's outputs can be stored in any note-taking app, offering durability and flexibility for organizing and editing handwritten content digitally.