AIR#44 - April 17, 2024
Good morning, AI aficionados! Today's edition of AIR: The AI Recon is like a treasure trove, packed with gems that shine a light on the incredible strides we're making in the world of artificial intelligence. From the NSA stepping up to bat with new guidelines aimed at fortifying AI system security against the ever-looming shadow of cyber threats, to DeepMind's latest marvel, "Aloha Unleashed," which is set to revolutionize the world of robotics with its low-cost, highly dexterous robots, we're on the cusp of witnessing AI reshape our digital and physical worlds in ways we've only dreamed of.
And if that's not enough to pique your interest, let's dive into the academic realm where the "ResearchAgent" is making waves. This groundbreaking tool uses AI to fast-track scientific discovery by generating novel research ideas from the vast ocean of scientific literature, showcasing the untapped potential of AI in accelerating our quest for knowledge and innovation. Meanwhile, across the pond, the UK is taking a firm stand against the misuse of AI, criminalizing the creation of sexually explicit deepfakes in a bold move to protect individuals from digital exploitation, signaling a growing awareness of the ethical implications of AI technology.
As you enjoy your morning coffee and explore these stories, let the breadth of today's selection remind you of the boundless possibilities and challenges that lie ahead in the AI odyssey. From enhancing global security and pushing the boundaries of robot dexterity to pioneering new frontiers in research and navigating the complex ethical landscape, each story is a testament to human ingenuity and the transformative power of artificial intelligence. So, let's embark on this journey together, exploring the marvels and moral questions of AI, one breakthrough at a time.
Business
NSA Releases Guidelines to Fortify AI System Security
NSA unveils new guidelines to boost AI system security, aiming to shield against growing cyber threats.
UK to Criminalize Creation of Sexually Explicit Deepfakes
UK to make creating sexually explicit deepfakes a crime, targeting misuse and protecting victims with fines or jail.
Time Magazine's Best Inventions Award to Humane's AI Pin Sparks Controversy
Time's Best Invention award to not-yet-shipped AI Pin raises eyebrows, especially as it flops post-launch. Owners' investment raises questions.
"AI Girlfriends: The Next Billion-Dollar Tech Revolution"
Tech exec predicts AI girlfriends to spark $1B industry as demand for virtual companions grows, reshaping digital intimacy.
TensorWave Bets on AMD Over Nvidia for AI Cloud Dominance
TensorWave bets on AMD's MI300X over Nvidia for AI cloud supremacy, aiming for cost-effective, high-performance AI acceleration with plans to scale up rapidly.
Microsoft's $1.5B Investment in G42 Boosts AI Development and Global Expansion
Microsoft's $1.5B investment in G42 aims to make UAE a global AI hub, expanding AI and cloud tech globally, with a focus on safety and innovation.
Rivos Secures Over $250M to Fuel AI Chip Development Amid Industry Boom
Rivos raises over $250M for AI chip development, tapping into the AI boom with a focus on cost-effective solutions for data analytics and generative AI.
Apple's iOS 18 to Feature On-Device AI for Enhanced Privacy
Apple's iOS 18 boosts privacy with on-device AI, reducing cloud reliance for Siri & other features, aiming for safer, offline processing.
AI Wearable Disaster: Humane AI Pin Slammed by Critics
Much-hyped Humane AI Pin faces backlash, critics slam the wearable disaster.
Snowflake Unveils Groundbreaking Text-Embedding Model for Efficient Data Retrieval
Snowflake's new Arctic embed models revolutionize data retrieval with unmatched efficiency and open-source accessibility, setting new benchmarks in AI.
Engineering
🔥 DeepMind's "Aloha Unleashed": Revolutionizing Robot Dexterity with AI
DeepMind's Aloha Unleashed brings low-cost, highly dexterous robots to the forefront, showcasing their skills in a stunning one-shot video.
🔥 [Paper] Megalodon: Surpassing Llama2 with Unlimited Context LLM Pretraining and Inference
Megalodon outperforms Llama2, offering unlimited context LLM pretraining and inference with groundbreaking efficiency and stability.
🔥 Understanding Vision Transformers: An Illustrated Guide
Dive into Vision Transformers: From pixel patches to image classification, unraveling the steps of how ViTs learn to see.
Supabase Launches AI Inference in Edge Functions: Run AI Models Easily
Supabase makes AI easy: Launches AI inference in Edge Functions, enabling quick AI model runs & embedding generation, slashing costs & boosting developer experience.
[GitHub] neurallambda's awesome-reasoning: Curated Datasets for AI Reasoning
Discover neurallambda's awesome-reasoning on GitHub: A goldmine of datasets for enhancing AI reasoning skills.
Mozilla's Llamafile: A Privacy-Focused LLM for Open Source Enthusiasts
Mozilla launches Llamafile, a privacy-centric LLM for open source fans, ensuring AI benefits without sacrificing privacy.
WizardLM Withdrawn for Missing Toxicity Tests, Promises Quick Return
WizardLM paused for missed toxicity tests, vows swift fix and return.
[GitHub] testup-ai-driver: Automate Web Testing with AI and Selenium
Testup-ai-driver on GitHub makes web testing easy with AI & Selenium, automating tests through natural language. Say goodbye to complex coding!
Creating AI Agents with No-Code: A Comprehensive Guide
Unlock AI agent creation without coding! Dive into no-code tools for AI with long-term memory & complex tasks automation. 🛠️🤖 #AIRecon
Carl Cortright Unveils flyflow-conversational-voice-v1: Ultra-Low-Latency Voice Model for Apps
Carl Cortright releases flyflow-conversational-voice-v1: A game-changer for app developers needing ultra-low-latency, realistic voice interfaces.
[GitHub] Termax: Transform Natural Language to Shell Commands with LLM and RAG, Featuring Data Security Tips
Termax on GitHub: Turn natural language into shell commands with LLM & RAG. Elevate terminal use with data safety in mind.
Academic
🔥 [Paper] ResearchAgent: Generating Research Ideas with LLMs from Scientific Literature
ResearchAgent uses AI to fast-track scientific discovery, generating novel research ideas by analyzing literature and refining them iteratively.
Debunking Bloomberg: ChatGPT Not Racist, But Flawed in Recruiting
Debunking Bloomberg: Analysis shows ChatGPT isn't racist but struggles with resume bias, overvaluing pedigree over potential.
Israel's AI Targeting Systems in Gaza Spark Fears of Innocent Casualties
Israel's AI systems in Gaza, "Lavender" & "Where's Daddy?", raise alarm over innocent casualties in accelerated kill chain operations.
[Paper] Zero-shot Building Age Estimation with GPT-4 Vision from Facade Images
GPT-4 Vision now predicts building ages from facades, showing promise despite a modest 39.69% accuracy and challenges with very old structures.