AIR#139 - July 24, 2024

Good morning, AI aficionados! As you sip your morning coffee, get ready to dive into today's edition of AIR: The AI Recon. Leading the buzz is Meta's monumental release of Llama 3.1, a game-changing open-source AI model available in versions as large as 405 billion parameters. This powerhouse not only aims to democratize AI but also to challenge closed systems like OpenAI's GPT-4. Whether you're a developer eager to fine-tune and deploy or a business leader eyeing cost-effective AI solutions, this story is a must-read.

But that's not all! Elon Musk's xAI Memphis Supercluster has gone live, boasting an eye-popping 100,000 Nvidia H100 GPUs. This supercluster aims to be the world's most powerful AI computing hub, setting the stage for unprecedented advancements in AI research and applications. If you're fascinated by the sheer scale and ambition of AI infrastructure, this is a story you won't want to miss. Meanwhile, OpenAI is shaking things up by reassigning their AI safety leader Mądry, signaling a shift in focus from catastrophic risk to fundamental research. This development could have far-reaching implications for the future of AI safety and ethics.

And for a touch of the uncanny, AI is making waves in the video game industry, but not in the way you'd expect. Automation is replacing jobs, causing a stir among developers and artists. This story delves into the anxieties and realities of an industry grappling with rapid technological change. Whether you're here for groundbreaking tech updates, ethical debates, or the latest industry buzz, today's edition is packed with stories that will both intrigue and challenge you. So, sit back, sip your coffee, and let's delve into the dynamic world of artificial intelligence together!

Business

🔥 Open Source AI Will Dominate: Meta's Llama 3.1 Release
Meta releases Llama 3.1, the first frontier-level open source AI model, aiming to make open source AI the industry standard.

Why Llama 3.1 Is a Game-Changer
Meta's Llama 3.1 is a game-changer: open-source, cost-effective, secure, and customizable, freeing companies from big AI providers.

Elon Musk's xAI Memphis Supercluster Goes Live with 100,000 Nvidia H100 GPUs
Elon Musk's xAI Memphis Supercluster goes live with up to 100,000 Nvidia H100 GPUs, aiming to be the world's most powerful AI cluster.

Trust at Scale: Auto-Evaluation for High-Stakes LLM Accuracy
Elicit uses LLM-based auto-evaluation with human oversight to ensure trustworthy, scalable, and flexible AI accuracy for scientific literature.

GM’s Cruise Abandons Origin Robotaxi, Takes $583M Charge
GM's Cruise ditches Origin robotaxi, takes $583M hit, shifts focus to next-gen Chevy Bolt for cost-effective, scalable autonomous future.

Elon Musk: Tesla to Deploy Humanoid Robots Next Year
Elon Musk says Tesla's humanoid robots, Optimus, will start factory tasks in 2025 and be available for sale by 2026.

Generative AI to Consume 1.5% of Global Power by 2029
Generative AI will consume 1.5% of global power by 2029, driven by the high energy demands of AI chips and data centers.

AI Companies Promised Self-Regulation: What’s Changed?
One year after AI firms pledged self-regulation, progress is mixed. Some advancements, but much work remains for true AI safety and transparency.

Meta releases Llama 3.1 open-source AI model to take on OpenAI
Meta launches Llama 3.1, the largest open-source AI model, to outdo OpenAI's GPT-4o, aiming to make Meta AI the top assistant by year-end.

Engineering

🔥 Introducing Llama 3.1: Fine-Tune, Distill, and Deploy Anywhere
Meet Llama 3.1: Fine-tune, distill, and deploy our open-source AI model anywhere. Available in 8B, 70B, and 405B versions.

🔥 Briefer: Multiplayer Cloud Notebooks with AI, SQL, and Scheduling
Briefer launches cloud-based notebooks with AI, SQL, and scheduling, offering collaborative data analysis and visualization tools.

[GitHub] DOM to Semantic Markdown for LLMs by Romansky
Convert HTML DOM to Semantic Markdown for LLMs with Romansky’s tool—preserves structure, metadata, and optimizes tokens.

Meta's Llama 3.1 405B: First Open-Weights GPT-4-Level AI Model
Meta's Llama 3.1 405B: first open-weights GPT-4-level AI model, free to download, challenging closed AI vendors like OpenAI.

Meta's Llama 3.1 405B: The Open-Source Giant Challenging GPT-4
Meta's Llama 3.1 405B challenges GPT-4 with open-source power, boasting 128k context and top eval scores for high-quality dialogue.

AI is Taking Over Jobs in the Video Game Industry
AI is replacing jobs in the video game industry, causing layoffs and anxiety among artists and developers fearing for their livelihoods.

Learnings from Building an AI Code Assistant with RAG Architecture
Built an AI Code Assistant with GPT-4o-mini, Langchain, Postgres, and pg_vector. It helps explore codebases by chatting with a virtual "senior developer."

[Paper] Domain-Aware Fine-Tuning of Foundation Models by Kaplan et al.
New study introduces Domino, a method for domain-aware fine-tuning of foundation models, enhancing adaptability to unseen domains.

[GitHub] Automate Factorio with AI Agents
Automate Factorio with AI! GitHub repo offers remote interfaces to control actions, manage resources, and streamline gameplay. 🚀

Iterating on an HTML Resume with GPT-4o via cURL
Using GPT-4o via cURL to refine an HTML resume. Automates feedback, improving clarity and precision. Script available on GitHub.

[GitHub] meta-llama/llama-agentic-system: Agentic Components for Llama Stack APIs
Meta-Llama's new repo offers agentic components for Llama Stack APIs, enabling multi-step reasoning and tool usage. 529 stars already! 🌟

Academic

OpenAI Removes AI Safety Leader Mądry, Former Ally of CEO Altman
OpenAI reassigns AI safety leader Mądry, shifting focus from catastrophic risk to fundamental research.

[Paper] Meta AI's Llama 3: A New Herd of Multilingual, Multimodal Models
Meta AI's Llama 3 models offer multilingual, multimodal capabilities, rivaling GPT-4 in various tasks. Public release includes safety features.

[Paper] ChatBCG: Can AI Read Your Slide Deck?
AI struggles to fully read and interpret complex or unlabeled charts in business slide decks, performing below human accuracy levels.

Is AI Winter Coming?
Doubts rise about AI's future as industry shifts from optimism to skepticism. Is AI Winter approaching? Engineers might be the solution.

A.I. Is Great at Poetry, Terrible at Math
AI excels at poetry but struggles with math, highlighting its limitations in precise, rules-based calculations.