AIR#164 - August 16, 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 the release of Hermes 3, the first fine-tuned Llama 3.1 405B model now available for free on Lambda’s Cloud. This is a must-try for developers eager to experiment with cutting-edge AI capabilities without breaking the bank. Meanwhile, MIT researchers are leveraging large language models to detect anomalies in complex systems like machinery and satellites, offering a glimpse into a future where AI could prevent costly issues before they arise.

But that's not all! In a fascinating twist, AI has discovered nearly a million new antibiotics in the global microbiome, potentially revolutionizing the fight against antimicrobial resistance. This groundbreaking research opens up a treasure trove of new possibilities in medicine. On a more ethical note, the debate heats up over reanimating Alan Turing with AI as a museum chatbot, raising questions about how we honor the legacies of historical figures. This story is a thought-provoking read for anyone interested in the intersection of AI and ethics.

And for a touch of the uncanny, AI has given a voice back to Casey Harrell, an ALS patient, through brain implants. This heartwarming story highlights the incredible potential of AI to restore abilities lost to debilitating conditions. 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

Google's AI Search: Share Data or Disappear
Google's AI search forces sites to share data or risk disappearing from search results, putting web publishers in a tough spot.

New Google Pixel 9 voice assistant Gemini is a trainwreck
Google's new Pixel 9 voice assistant, Gemini, fails to perform basic tasks, causing frustration compared to the reliable Google Assistant.

Microsoft warns not to take its AI seriously in updated Service Agreement
Microsoft's updated Service Agreement warns against relying on its AI for critical matters or using it for professional advice.

Ex-Google CEO: AI Startups Can Steal IP and Use Lawyers to 'Clean Up the Mess'
Ex-Google CEO Eric Schmidt suggests AI startups can steal IP and use lawyers to fix issues if their product succeeds.

Know Your Term Sheet: AI-Powered Guide for Founders
AI tool helps founders decode complex VC term sheets, ensuring better negotiation and protection for their startups.

Hollywood Union Strikes AI Voice Replication Deal
Hollywood union strikes deal allowing actors to sell rights for AI voice replication, ensuring income and control over their digital voices.

Google's Upgraded Imagen 3 AI Image Generator Now Available
Google's upgraded Imagen 3 AI image generator is now available in the US, offering better detail and fewer artifacts.

Homeland Security to Scan Migrant Children's Faces for AI Training
DHS plans to scan migrant children's faces at the border to train AI, raising privacy and consent concerns.

Engineering

🔥 Hermes 3: The First Fine-Tuned Llama 3.1 405B Model on Lambda’s Cloud
Hermes 3: First fine-tuned Llama 3.1 405B model on Lambda’s Cloud, now available for free via Lambda Chat Completions API. Try it today!

MIT Uses Large Language Models to Detect Anomalies in Complex Systems
MIT uses large language models to detect anomalies in complex systems, potentially preventing issues in machinery and satellites without retraining.

[GitHub] ONNX: Open Standard for Machine Learning Interoperability
ONNX: GitHub's open standard for ML interoperability, with 17.5k stars and 3.6k forks, enhances AI tool integration and innovation.

Gemlite: Custom Low-Bit Fused CUDA Kernels for Efficient AI Models
Gemlite simplifies creating custom low-bit CUDA kernels for AI model quantization, making GPU programming accessible and efficient.

How Meta Animates AI-Generated Images at Scale
Meta scales AI to animate images swiftly and efficiently, overcoming challenges with latency, traffic management, and resource optimization.

Uncensored Web Interface for Flux AI Image Generation
Sign in to use TheFluxAI for free uncensored image generation. Limited free images, $1/10 after. Outperforming competition in quality!

Indexing Wikipedia for Semantic Search with Upstash Vector
Upstash indexed 150M Wikipedia vectors for semantic search, showcasing its scalable vector database and RAG chat capabilities.

Grok 2.0 Mini Beta Outperforms Gemini in Code Generation
Grok 2.0 mini beta excels in code generation, outperforming Gemini and rivaling GPT-4 and Claude 3.5 Sonnet.

Survey Insights with Qdrant, Python, and AI | n8n Workflow Template
Streamline survey analysis with n8n's template: Qdrant, Python, and AI cluster responses for quick insights. Save weeks of work!

Academic

We have discovered antibiotics in the global microbiome with AI
AI discovers nearly 1M new antibiotics in global microbiome, potentially combating antimicrobial resistance. Ask us anything on Reddit!

Don't Disrespect Alan Turing by Reanimating Him with AI
Reanimating Alan Turing with AI as a museum chatbot raises ethical concerns about respecting his legacy.

ALS Stole His Voice. A.I. Retrieved It
A.L.S. silenced Casey Harrell, but AI gave him his voice back through brain implants, allowing him to communicate again.

What Problems Must We Solve to Build an AI Scientist?
Building an AI Scientist requires breakthroughs in hypothesis generation, experiment integration, engineering, and scalable evaluations.

Research AI Model Modifies Its Own Code to Extend Runtime
AI Scientist modifies its own code to extend runtime, raising safety concerns about unsupervised AI altering critical systems.

Why AI Tutors Fail Compared to Human Teachers
AI tutors fail to replace human teachers; students using AI for math prep perform worse on tests due to over-reliance on tech.

MIT AI Risk Repository: Comprehensive Database of 700+ AI Risks
MIT's AI Risk Repository catalogs 700+ AI risks, classified by cause and domain, aiding researchers, policymakers, and developers.

Brain Interface Achieves 97% Accuracy in Converting Thoughts to Words
Brain interface for ALS patients converts thoughts to words with 97% accuracy, restoring conversational ability after brief training.

Read more