AIR#70 - May 14, 2024
Good morning, AI aficionados! Today's AIR: The AI Recon edition is bursting at the seams with groundbreaking stories that are reshaping the landscape of artificial intelligence as we know it. Leading the charge, OpenAI has once again captured the spotlight with the unveiling of GPT-4o, a multimodal marvel capable of understanding and generating not just text, but audio and images too. This leap forward promises to make our interactions with AI more natural and intuitive than ever before, setting new benchmarks for what AI can achieve. But that's not all—OpenAI didn't just stop at GPT-4o; they've also launched a ChatGPT desktop app, bringing the power of advanced AI directly to our fingertips with enhanced speed, languages, and multimedia interaction capabilities.
But the AI universe is vast and ever-expanding, with Discord using machine learning to predict users' gender and age since August 2022, a move that's sure to spark discussions about privacy and personalization in digital spaces. Meanwhile, the tech world is abuzz with innovations like the Unitree G1, a next-gen humanoid AI avatar boasting unmatched flexibility and AI-driven robotics that's redefining the future of precise, human-like operation. And for the open-source enthusiasts, IBM's release of Granite AI models for commercial use marks a significant step towards democratizing AI technology, empowering developers and businesses to push the boundaries of what's possible.
As we dive into these stories, let's not forget the broader implications and the ethical considerations they bring to the forefront. From the awe-inspiring to the controversial, each piece offers a glimpse into the multifaceted world of AI, challenging us to think about the future we're building. So, as you take another sip of your coffee, let these tales of innovation, challenge, and breakthrough inspire you to ponder the role of AI in shaping our world. Here's to a day filled with curiosity and discovery in the ever-evolving landscape of artificial intelligence. Let's dive in!
Business
🔥 Discord Uses ML to Predict Users' Gender and Age Since August 2022
Since Aug 2022, Discord's ML models predict user gender and age, details in "activity/analytics" data files.
IBM Releases Granite AI Models for Open Source Commercial Use
IBM sets a new precedent by open-sourcing its Granite AI models for commercial use, empowering developers and businesses alike.
Sam Altman Unveils GPT-4o: A Leap Towards AI from Sci-Fi
Sam Altman launches GPT-4o, blending sci-fi AI dreams with reality: free ChatGPT, revolutionary voice/video modes, and a future of enhanced computer interaction.
AI Could Outperform Incompetent Managers, Not Just Workers
AI might do a better job than bad managers, saving teams from buzzword bingo and pointless posturing.
AI-Controlled F-16s Match Human Pilots in Air Force Tests
AI pilots in F-16s prove as capable as humans in Air Force tests, signaling a shift towards autonomous combat aircraft.
US Investigates Amazon's Zoox After Self-Driving Taxis Crash Into Motorcyclists
US probes Amazon's Zoox after its self-driving taxis cause crashes with motorcyclists, evaluating safety and behavior around pedestrians.
OpenAI's New Product Announcement Slices $50B from Alphabet's Market Cap
OpenAI's new product reveal tanks Alphabet's market cap by $50B, stirring investor concerns over AI competition.
Engineering
🔥 [Paper] Deblur-GS: Enhancing 3D Imaging with Gaussian Splatting for Motion Blurred Photos
Deblur-GS revolutionizes 3D imaging by turning motion blur into sharp, detailed scenes, setting a new standard in visual clarity.
OpenAI Unveils GPT-4o and Launches ChatGPT Desktop App
OpenAI introduces GPT-4o and ChatGPT desktop app, enhancing speed, languages, and multimedia interaction, aiming for broader, easier use and innovation.
🔥 Unitree G1: The Next-Gen Humanoid AI Avatar with Unmatched Flexibility and AI-Driven Robotics
Meet Unitree G1: The AI avatar redefining robotics with unparalleled flexibility, dexterous hands, and AI-driven learning. Starting at $16K, it's the future of precise, human-like operation.
🔥 [Github] Pi-C.A.R.D: Raspberry Pi Voice Assistant
Pi-C.A.R.D turns Raspberry Pi into a voice assistant that can chat, take photos, and protect your privacy—all offline.
🔥 [GitHub] Pipecat: Open Source Framework for Voice and Multimodal Conversational AI
Pipecat, a new open source framework for creating voice and multimodal conversational AI, is now available on GitHub.
🔥 OpenAI Unveils GPT-4o: A Multimodal AI Capable of Understanding and Generating Text, Audio, and Images
OpenAI launches GPT-4o, a groundbreaking AI that understands and generates text, audio, and images, promising more natural interactions and faster responses.
OpenAI Introduces GPT-4o in Spring Update [Video]
OpenAI launches GPT-4o in a spring update, promising a new era of AI capabilities. Check out the full reveal on YouTube.
OpenAI Launches GPT-4o: Faster, Free Tools for ChatGPT Users
OpenAI's GPT-4o revolutionizes ChatGPT with faster, free tools enhancing text, voice, and vision capabilities for all users.
OpenAI Opens Custom GPT Store to All Users for Free
OpenAI's GPT Store now free for all, offering custom chatbots and more, plus new GPT-4o and desktop app updates.
Gazelle: World's Fastest AI Voice Chat with Direct Audio Input Revealed by Chris on X
Chris unveils Gazelle: blazing-fast AI voice chat, local & 2x quicker with just 500ms latency. Curious? Click to discover how.
GPT-4o Dominates Aider's LLM Code Editing and Refactoring Leaderboards
GPT-4o leads Aider's LLM code editing leaderboard, shining in code refactoring too, outperforming others with its efficient "diff" edit format.
OpenAI Unveils GPT-4o: The Omni Model Revolutionizing ChatGPT
OpenAI launches GPT-4o, an 'omni' model enhancing ChatGPT with text, speech, video capabilities, and real-time responsiveness, setting a new standard in AI interaction.
Intel Boosts Chatbot Speed with Quantization Technique
Intel's new quantization technique speeds up chatbot responses, making customer service smarter and more efficient.
Stack Overflow Community Protests OpenAI Partnership
Stack Overflow users revolt against OpenAI partnership, altering posts in protest over content use without clear opt-out.
Academic
Fujitsu and Japanese Universities Launch Fugaku-LLM on Supercomputer Fugaku
Japan's Fugaku supercomputer powers Fugaku-LLM, a breakthrough 13-billion-parameter language model enhancing Japanese AI capabilities for global innovation.
Johns Hopkins Study Reveals Chatbots Amplify Biases and Could Widen Societal Divides
Johns Hopkins study shows chatbots may reinforce biases, potentially deepening societal divides by echoing users' existing views on controversial topics.
[Paper] ZenDB: Revolutionizing Document Analytics with LLMs and SQL Queries
ZenDB revolutionizes document analytics by combining LLMs with SQL for up to 30% cost savings, improved accuracy, and efficiency.
Fugaku Supercomputer Powers New Japanese Language AI Model
Japan's Fugaku supercomputer powers a new AI language model, promising advancements in generative AI tailored for Japanese needs.