Welcome to Flash AI Digest, your biweekly update on the latest in artificial intelligence! Each edition brings you:

• AI News: Key developments, breakthroughs, and trends shaping the AI landscape.
• Image Generation Spotlight: Creative and cutting-edge visual innovations powered by AI.
• Highlighted Paper: A featured research paper, offering deep insights into the future of AI technology.

Stay informed, inspired, and ahead of the curve with Flash AI Digest!

🗞️ News

OpenAI announces o3, achieves 87.5% on ARC-AGI, outperforms prior models

On the final day of its ’12 Days of Christmas’, OpenAI releases o3, the latest model in its reasoning series, built to solve complex, unseen problems. It moves beyond pattern-matching by generating solutions on the fly using a hybrid neural-symbolic framework.

Use Ollama with any GGUF Model on Hugging Face Hub

Ollama is an application based on llama.cpp to interact with LLMs directly through your computer. You can use any GGUF quants created by the community on Hugging Face directly with Ollama, without creating a new Modelfile

GitHub Copilot Now Free in VS Code

GitHub has announced that Copilot is now free for all users in Visual Studio Code. By signing in with a personal GitHub account, developers can access 2,000 code completions and 50 chat messages per month. This free tier offers a choice between Anthropic’s Claude 3.5 Sonnet and OpenAI’s GPT-4o models.

1–800-CHATGPT

OpenAI has made voice and text communication possible with ChatGPT. In the US, you can call 1–800-CHATGPT for 15 free minutes a month or send a WhatsApp message from anywhere in the world.

Google launches Veo 2, 4K video model, and enhanced image model globally.

Google launches three new tools: Veo 2 for video generation, Imagen 3 for image creation, and Whisk for remixing visuals. These tools provide high-resolution video, accurate image generation, and creative control over visuals.

🖌️ AI Image Generation Spotlight

Model: DALL-E
Prompt: “A cozy Christmas-themed indoor scene featuring a cheerful young woman sitting at a wooden table. She has shoulder-length dark hair, big expressive eyes, and a warm smile. She is wearing a bright red hoodie and a festive Santa hat with a white pom-pom. In front of her is an open laptop glowing softly, displaying the text ‘See you in 2025’ in a decorative font with snowflake accents. The background includes a large window showing a clear winter night sky with a full moon and snow-covered trees. Inside, there are Christmas decorations, including a lit Christmas tree adorned with ornaments, strings of fairy lights, and warm, ambient lighting creating a magical holiday atmosphere. Books and small holiday items are on the table, enhancing the cozy and festive vibe.”

🎓Highlighted Paper

Byte Latent Transformer: Patches Scale Better Than Tokens

If you wanna read older editions, you can find them on the publication page

Flash AI Digest

If you find this post interesting, kindly consider liking and sharing it with others, allowing more people to come across it.

If you’re curious about technology and software development, I invite you to explore my other posts. You’ll find a range of topics that delve into the world of coding, artificial intelligence, and the latest tech trends. Whether you’re a many years professional developer or just starting, there’s something here for everyone.

References

• https://openai.com/12-days/?day=12
• https://huggingface.co/docs/hub/en/ollama
• https://code.visualstudio.com/blogs/2024/12/18/free-github-copilot
  https://help.openai.com/en/articles/10193193-1-800-chatgpt-calling-and-messaging-chatgpt-with-your-phone
• https://blog.google/technology/google-labs/video-image-generation-update-december-2024/
• https://creator.nightcafe.studio/creation/QzryQqbS1a9l6sFmLMLt?ru=GeyikGames
• https://arxiv.org/abs/2412.09871v1

📰 AI News Biweekly was originally published in Flash AI Digest on Medium, where people are continuing the conversation by highlighting and responding to this story.

Welcome to Flash AI Digest, your biweekly update on the latest in artificial intelligence! Each edition brings you:

• AI News: Key developments, breakthroughs, and trends shaping the AI landscape.
• Image Generation Spotlight: Creative and cutting-edge visual innovations powered by AI.
• Highlighted Paper: A featured research paper, offering deep insights into the future of AI technology.

Stay informed, inspired, and ahead of the curve with Flash AI Digest!

🗞️ News

Harvard University Releases 1 Mililion Public Domain Books For AI Training

Harvard, with Microsoft and OpenAI, released a dataset of 1M public-domain books from Google Books to boost AI training, democratize access to quality data, and support smaller AI developers alongside licensed
content.

Meta Launched Llama 3.3 70B Instruct Model

Meta has launched the Llama 3.3 70B Instruct model, a 70-billion parameter language model optimized for multilingual dialogue. It excels in reasoning, math, and general knowledge, surpassing many models. It is available on Hugging Face.

Microsoft Introduces PHI-4: Groundbreaking AI For Complex Tasks

Microsoft has introduced Phi-4, a 14-billion parameter small language model designed for complex reasoning tasks, including mathematics. Phi-4 outperforms larger models like Gemini Pro 1.5 in math-related benchmarks.

OpenAI Launches Sora: The Future of Video Editing

OpenAI launches Sora, a cutting-edge video editor designed to simplify and elevate content creation. With AI-powered tools, creators can edit videos effortlessly and produce stunning results. Explore the future of video editing with Sora!

Google Develops AI to Autonomously Perform Computer Tasks

Project Jarvis, Google’s new AI, aims to handle online tasks like research and bookings by controlling web browsers. Powered by the upcoming Gemini model, it previews in December, enhancing user productivity through task automation.

🖌️ AI Image Generation Spotlight

Model: Flux Schnell
Prompt: “A cartoon Santa Claus, wearing a red hat and coat with white fur trim, and a pair of jeans, plays a banjo with a joyful expression, sitting on a stool near an outdoor cafe at night, surrounded by twinkling string lights and steam rising from cups of hot cocoa, in the style of Mark Brooks’ dynamic lines and Dan Mumford’s vibrant colors, with a perfect blend of smooth textures and bold graphic elements, reminiscent of classic comic book art.”

🎓Highlighted Paper

3D-LLM: Injecting the 3D World into Large Language Models

If you wanna read older editions, you can find them on the publication page

Flash AI Digest

If you find this post interesting, kindly consider liking and sharing it with others, allowing more people to come across it.

If you’re curious about technology and software development, I invite you to explore my other posts. You’ll find a range of topics that delve into the world of coding, artificial intelligence, and the latest tech trends. Whether you’re a many years professional developer or just starting, there’s something here for everyone.

References

• Harvard Is Releasing a Massive Free AI Training Dataset Funded by OpenAI and Microsoft
• meta-llama/Llama-3.3-70B-Instruct · Hugging Face
• Introducing Phi-4: Microsoft's Newest Small Language Model Specializing in Complex Reasoning | Microsoft Community Hub

https://openai.com/index/sora-is-here/

Google's Project Jarvis: A New AI Agent That Could Transform Everyday Web Browsing

https://creator.nightcafe.studio/creation/iXYd72oCwrL7if4F1y42/cartoon-santa-playing-banjo-at-night-near-an-outdoor-cafe

3D-LLM: Injecting the 3D World into Large Language Models

📰 AI News Biweekly was originally published in Flash AI Digest on Medium, where people are continuing the conversation by highlighting and responding to this story.

Welcome to Flash AI Digest, your biweekly update on the latest in artificial intelligence! Each edition brings you:

• AI News: Key developments, breakthroughs, and trends shaping the AI landscape.
• Image Generation Spotlight: Creative and cutting-edge visual innovations powered by AI.
• Highlighted Paper: A featured research paper, offering deep insights into the future of AI technology.

Stay informed, inspired, and ahead of the curve with Flash AI Digest!

🗞️ News

Perplexica: AI Powered Search Engine

Perplexica is an open-source AI-powered searching tool or an AI-powered search engine that goes deep into the internet to find answers. Inspired by Perplexity AI, it’s an open-source option that not just searches the web but understands your questions.

Andrew Ng’s AISuite Lets You Test Multiple LLMs by Changing One String

AISuite is an open-source Python package for integrating multiple LLMs. Developers switch LLMs by specifying <provider>:<model-name>. It supports various providers and focuses on chat completions.

Android 16 and Agent Gemini

Google will position its own artificial intelligence model, codenamed “Gemini”, as an “agent” in version 16 of the Android operating system. This means that Gemini will be able to take action and extract information on your behalf in your installed applications

World’s Most Flexible Sound Machine Debuts

Using text and audio as inputs, a new generative AI model from NVIDIA can create any combination of music, voices and sounds.

Who is David Mayer and Why Won’t ChatGPT Say His Name?

David de Rothschild, adventurer and eco-activist, charts a unique path beyond his banking lineage. Despite his fame, ChatGPT avoids his name, raising questions of a “digital gag order” or a strange glitch in AI’s programming

🖌️ AI Image Generation Spotlight

Model: Google Imagen
Prompt: “hyper-realistic image of a slimey wacky snail holding an envelope in its mouth, set in a dimly lit, soft, natural environment with a hint of mist. The snail should have exaggerated, cartoonish features, with large, expressive eyes that make it look curious and endearing. Its shell should be textured with fine details and subtle highlights. Add small water droplets on the snail’s body and eyes, enhancing its wet, slimy look. The background should be softly blurred city street, giving a sense of depth and mystery”

🎓Highlighted Paper

Star Attention: Efficient LLM Inference over Long Sequences

[2411.17116] Star Attention: Efficient LLM Inference over Long Sequences

If you wanna read older editions, you can find them on the publication page

Flash AI Digest

If you find this post interesting, kindly consider liking and sharing it with others, allowing more people to come across it.

If you’re curious about technology and software development, I invite you to explore my other posts. You’ll find a range of topics that delve into the world of coding, artificial intelligence, and the latest tech trends. Whether you’re a many years professional developer or just starting, there’s something here for everyone.

References

GitHub — ItzCrazyKns/Perplexica: Perplexica is an AI-powered search engine. It is an Open source alternative to Perplexity AI

GitHub — andrewyng/aisuite: Simple, unified interface to multiple Generative AI providers

Android 16 may give Gemini new power over your apps — Android Authority

Fugatto, World’s Most Flexible Sound Machine, Debuts | NVIDIA Blog

What’s going on with ChatGPT and the name ‘David Mayer’? | Mashable

Snail mail — Snail mail

[2411.17116] Star Attention: Efficient LLM Inference over Long Sequences

📰 AI News Biweekly was originally published in Flash AI Digest on Medium, where people are continuing the conversation by highlighting and responding to this story.

Welcome to Flash AI Digest, your biweekly update on the latest in artificial intelligence! Each edition brings you:

• AI News: Key developments, breakthroughs, and trends shaping the AI landscape.
• Image Generation Spotlight: Creative and cutting-edge visual innovations powered by AI.
• Highlighted Paper: A featured research paper, offering deep insights into the future of AI technology.

Stay informed, inspired, and ahead of the curve with Flash AI Digest!

🗞️ News

Google Launches Gemini AI Bot App For IOS Users

Google has introduced its ChatGPT rival, “Gemini,” as a dedicated app for iOS. The AI bot aims to revolutionize how users interact with technology, providing a seamless and intelligent conversational experience

Github Copilot Will Support Models From Anthropic, Google and OpenAI

GitHub enhances its Copilot tool with AI models from Anthropic, Google, and OpenAI, allowing developers to choose the best fit for coding tasks. The new ‘Spark’ tool also simplifies web app development using natural language prompts.

AI-DA’S Alan Turing Portrait Sells For $1.1 Million

The AI-powered humanoid robot artist Ai-Da’s portrait of Alan Turing, the father of artificial intelligence, was sold for $1.1 million at a Sotheby’s auction, reaching 10 times its estimated value.

Introuducing Yolo11: Breakthroughs in Real-Time Object Detection

YOLO11, the latest in the Ultralytics YOLO series, excels in real-time object detection with unmatched accuracy, speed, and efficiency. With its improved architecture and training methods, YOLO11 suits diverse computer vision tasks.

ChatGPT Introduces Advanced Developer Integration

ChatGPT for macOS now integrates with VS Code, Xcode, and Terminal, offering real-time coding aid. Exclusive to ChatGPT Plus/Teams users, wider support is planned. An advanced version launches in Jan 2025 to expand its tools.

🖌️ AI Image Generation Spotlight

Model: Dall-E
Prompt: “Stunning image of a cute adorable mummy listening to music, moody, grainy, noisy, concept art, Meaningful Visual Art, Detailed Painting, Digital Illustration, Unreal Engine 5, 32k maximalist, hyperdetailed fantasy art, 3d digital art, sharp focus, masterpiece, fine art.”

🎓Highlighted Paper

1-bit AI Infra: Part 1.1, Fast and Lossless BitNet b1.58 Inference on CPUs

[2410.16144] 1-bit AI Infra: Part 1.1, Fast and Lossless BitNet b1.58 Inference on CPUs

If you wanna read older editions, you can find them on the publication page

Flash AI Digest

If you find this post interesting, kindly consider liking and sharing it with others, allowing more people to come across it.

If you’re curious about technology and software development, I invite you to explore my other posts. You’ll find a range of topics that delve into the world of coding, artificial intelligence, and the latest tech trends. Whether you’re a many years professional developer or just starting, there’s something here for everyone.

References

Download the Gemini app for iPhone today

GitHub Copilot will support models from Anthropic, Google, and OpenAI — The Verge

AI Robot Artist’s Turing Portrait Fetches $1.1 Million at Sotheby’s — Decrypt

YOLO11 🚀 NEW — Ultralytics YOLO Docs

Work with Apps on macOS | OpenAI Help Center

Wrap music — AI Generated Artwork — NightCafe Creator

[2410.16144] 1-bit AI Infra: Part 1.1, Fast and Lossless BitNet b1.58 Inference on CPUs

📰 AI News Biweekly was originally published in Flash AI Digest on Medium, where people are continuing the conversation by highlighting and responding to this story.

Welcome to Flash AI Digest, your biweekly update on the latest in artificial intelligence! Each edition brings you:

• AI News: Key developments, breakthroughs, and trends shaping the AI landscape.
• Image Generation Spotlight: Creative and cutting-edge visual innovations powered by AI.
• Highlighted Paper: A featured research paper, offering deep insights into the future of AI technology.

Stay informed, inspired, and ahead of the curve with Flash AI Digest!

🗞️ News

GameGen AI: Simplifying Game Asset Creation for Devs

Google Research, in collaboration with Tel Aviv University, has introduced GameNGen, the first real-time neural model-powered game engine. This innovative system simulates gameplay environments, such as the classic DOOM, at over 20 frames per second on a single TPU. Human testers struggled to distinguish between real and AI-simulated footage. The engine operates by training a reinforcement learning (RL) agent to play, using its actions and observations as data for a diffusion model. This model predicts future frames, enabling seamless long-term gameplay simulation with high-quality visuals.

GitHub integrates OpenAI’s new models

OpenAI has introduced a preview of its new o1 series AI models, including o1-preview and o1-mini, available for developers through GitHub Copilot and Models on Azure. These models, designed for advanced reasoning, enable deeper understanding of code constraints and edge cases, leading to more efficient results. Users can toggle between models in Visual Studio Code’s Copilot Chat to handle tasks ranging from simple API explanations to complex algorithm design. The models are also available for testing in GitHub Models, with opportunities for future integration into custom applications.

NotebookLM Summarizes YouTube Videos and Audio with New Feature

Google’s AI-powered note-taking tool, NotebookLM, has received an update allowing it to summarize YouTube videos and audio files. This new feature enhances the tool’s ability to assist users in processing multimedia content efficiently

Amazon Partners with Anthropic’s Claude AI for Alexa’s Next Evolution

Amazon is set to launch a revamped version of Alexa, powered by Anthropic’s Claude AI, in October 2024. This upgrade comes after Amazon’s in-house AI models struggled to meet performance benchmarks. The “Remarkable” version of Alexa will use generative AI to handle complex tasks like conversations, shopping advice, and home automation, while a free “Classic” version remains available. Amazon plans to charge $5 to $10 monthly for the new service, aiming to boost revenue. With Alexa lagging behind competitors like ChatGPT, this move is seen as critical for the division’s future success.

Meta Launched, Llama 3.2 at Connect 2024

Llama 3.2 has officially launched, marking an advancement in AI capabilities. Released by Hugging Face in collaboration with Meta, Llama 3.2 introduces both multimodal and text-only models, including the highly anticipated Llama 3.2 Vision. The Vision model supports visual reasoning, document question answering, and image-text retrieval with real-world applications like infographics and document processing. With sizes ranging from 1B to 90B parameters, these models offer scalable solutions for both consumer GPUs and large-scale enterprises. Llama Guard 3 adds enhanced input classification, ensuring safe AI interactions.

🖌️ AI Image Generation Spotlight

Model: RealVisXL
Prompt: “Hyper realistic cute teddy bear sitting in a leaf boat made out of leaves, in a puddle, alone on rainy day. Raincoat. Hyperdetailed, realism, blurry background, incredible composition, dynamic lighting, meticulously composed concept art, masterpiece, mystic fog, digital illustration, Volumetric lighting”

🎓Highlighted Paper

Gandhi, Kanishk, et al. “Human-like Affective Cognition in Foundation Models.” arXiv preprint arXiv:2409.11733 (2024).

Human-like Affective Cognition in Foundation Models

If you wanna read older editions, you can find them on the publication page

Flash AI Digest

If you find this post interesting, kindly consider liking and sharing it with others, allowing more people to come across it.

If you’re curious about technology and software development, I invite you to explore my other posts. You’ll find a range of topics that delve into the world of coding, artificial intelligence, and the latest tech trends. Whether you’re a many years professional developer or just starting, there’s something here for everyone.

References

• Llama can now see and run on your device - welcome Llama 3.2
• Ask Claude: Amazon turns to Anthropic's AI for Alexa revamp
• NotebookLM adds audio and YouTube support, plus easier sharing of Audio Overviews
• Try out OpenAI o1 in GitHub Copilot and Models
• GameNGen

📰 AI News Biweekly was originally published in Flash AI Digest on Medium, where people are continuing the conversation by highlighting and responding to this story.

Welcome to Flash AI Digest, your biweekly update on the latest in artificial intelligence! Each edition brings you:

• AI News: Key developments, breakthroughs, and trends shaping the AI landscape.
• Image Generation Spotlight: Creative and cutting-edge visual innovations powered by AI.
• Highlighted Paper: A featured research paper, offering deep insights into the future of AI technology.

Stay informed, inspired, and ahead of the curve with Flash AI Digest!

🗞️ News

Apple Intelligence: AI-Powered Tools for Enhanced Productivity and Expression

Apple Intelligence introduces a suite of AI-powered features designed to make writing, communication, and content creation more effortless across iPhone, iPad, and Mac. Writing Tools assist with summarizing text, drafting replies, and adjusting tone. Image Playground lets users create custom visuals, while Focus modes prioritize notifications based on importance. Siri is enhanced with better contextual understanding and integration with ChatGPT. All these features are built with privacy at the core, ensuring personal data remains secure and accessible only to the user.

OpenAI Releases o1-preview & o1-mini Language Models

OpenAI introduces the o1 series, new AI models designed for advanced reasoning tasks in science, coding, and math. These models excel at solving complex problems, outperforming previous versions in fields like physics, chemistry, and programming. The o1-preview is now available for ChatGPT and API users, with o1-mini offering a faster, cost-effective option. These models have improved safety protocols and are being evaluated in collaboration with U.S. and U.K. AI Safety Institutes. Though early in development, future updates will include broader capabilities like browsing and file uploads.

Google Integrates Gemini AI into Chrome Address Bar

Google has integrated the Gemini artificial intelligence chatbot into Chrome’s address bar. You can interact with Gemini using the command “@gemini”.

Microsoft Copilot Expands: AI-Enhanced Excel, PowerPoint, and More

Microsoft 365 Copilot has introduced powerful new features across its suite of applications, most notably in Excel, where it assists users with formulas, Python-based data analysis, and text-based data searches. In PowerPoint, it can create entire presentations from prompts, incorporating AI-generated images. New features also enhance Teams, Outlook, and Word, streamlining tasks like meeting summaries, email prioritization, and document drafting. Copilot Pages offers collaborative capabilities, and customizable AI Agents help automate tasks. These updates aim to make Microsoft 365 more efficient for both individual and enterprise users.

YouTube Integrates AI to Help Creators with Ideas, Titles, and Full Video Production

YouTube is enhancing its platform with new AI-powered features to streamline the creative process. The Inspiration Tab in YouTube Studio suggests video ideas, titles, thumbnails, and even outlines for creators, acting as a comprehensive brainstorming tool. The integration of DeepMind’s Veo in YouTube Shorts allows creators to generate AI-driven videos and backgrounds through the new Dream Screen feature, though these clips are limited to six seconds to avoid awkward AI results. Additionally, YouTube’s auto-dubbing feature will support more languages, and creators can interact with fans through new AI tools in the Communities section. As YouTube embraces AI to make content creation easier, there are concerns that it may lead to a flood of similar-looking AI-generated videos, but it also offers creators an opportunity to produce content faster and more efficiently.

🖌️ AI Image Generation Spotlight

Model: Dall-E
Prompt: “Stunning image of a cute adorable wacky tired looking bunny with a cloud above his head that is raining, moody, grainy, noisy, concept art, Meaningful Visual Art, Detailed Painting, Digital Illustration, Unreal Engine 5, 32k maximalist, hyperdetailed fantasy art, 3d digital art, sharp focus, masterpiece, fine art”

🎓Highlighted Paper

Jim Waldo and Soline Boussard. 2024. GPTs and Hallucination: Why do large language models hallucinate? Queue 22, 4, Pages 10 (July/August 2024), 15 pages.

GPTs and Hallucination: Why do large language models hallucinate?: Queue: Vol 22, No 4

If you find this post interesting, kindly consider liking and sharing it with others, allowing more people to come across it.

If you’re curious about technology and software development, I invite you to explore my other posts. You’ll find a range of topics that delve into the world of coding, artificial intelligence, and the latest tech trends. Whether you’re a many years professional developer or just starting, there’s something here for everyone.

References

• Apple Intelligence and Siri
• Use the Gemini web app to get answers in Chrome
• Microsoft Copilot can now help with Excel formulas, make PowerPoints, and more
• YouTube will use AI to generate ideas, titles, and even full videos
• GPTs and Hallucination: Why do large language models hallucinate?: Queue: Vol 22, No 4

https://openai.com/index/introducing-openai-o1-preview/

📰 AI News Biweekly was originally published in Flash AI Digest on Medium, where people are continuing the conversation by highlighting and responding to this story.