AIR#234 - Claude 3.5 Takes Center Stage: New Sonnet & Lithium's Future in Arkansas!

Hey there!

Here's the latest AI news for today. Enjoy!

Today's top stories

šŸ”„ Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku
Anthropic announces Claude 3.5 Sonnet and Haiku, featuring improved coding skills and a new computer use capability in beta.

šŸ”„ USGS uses machine learning to show large lithium potential in Arkansas
USGS reveals Arkansas' Smackover Formation holds 5-19 million tons of lithium, potentially meeting future EV battery demand.

šŸ”„ Guide to Fine-Tuning LLMs
The guide reviews fine-tuning techniques for LLMs, covering methodologies, challenges, and best practices for effective deployment.

StabilityAI releases Stable Diffusion 3.5
StabilityAI launches Stable Diffusion 3.5, enhancing realism, prompt adherence, and customization for AI image generation.

Show HN: Steiner ā€“ An open-source reasoning model inspired by OpenAI o1
Yichao Ji's Steiner model aims to replicate OpenAI o1's reasoning capabilities but faces challenges in inference-time scaling.

IBM's new SWE agents for developers
IBM's SWE-Agent 1.0 uses open LLMs to autonomously resolve GitHub issues, streamlining bug fixes for developers efficiently.

Show HN: AI Image Upscaler and Photo Enhancer with up to 10x resolution boost
AI image upscaler enhances photos with up to 10x resolution boost but faced timeout issues during navigation.

Transformers Utilization in Chart Understanding: A Review of Advances and Future
This review explores transformer advancements in chart understanding, highlighting frameworks, challenges, and future directions.

RunwayML releases Act One: obsoleting traditional motion capture
RunwayML's Act-One revolutionizes character animation, enabling expressive performances from simple video inputs without traditional motion capture.

X changed its terms of service to let its AI train on everyone's posts
X's new terms allow AI to train on all user posts, sparking outrage over privacy and content ownership.

Transformers.js 3.0 Released with WebGPU Support
Transformers.js 3.0 launched with WebGPU support, enhancing performance and adding new models, tasks, and quantization options.

Stable Diffusion 3.5
Stable Diffusion 3.5 launches with customizable models for diverse users, enhancing image quality and prompt adherence on consumer hardware.

Anthropic Claude 3.5 Sonnet and Haiku
Anthropic introduces Claude 3.5, enhancing AI's creative capabilities with sonnets and haikus.

Show HN: AndyNote ā€“ Chat-style transcripts with real-time speaker detection
AndyNote offers live transcription with real-time speaker detection for privacy-focused conversations.

How 15 Top LLMs perform on classification: accuracy vs. cost breakdown
A study evaluates 15 LLMs for classification tasks, comparing accuracy and cost, revealing trade-offs and top performers.

Mochi 1 preview. A new SOTA in open-source video generation. Apache 2.0
Mochi 1 sets a new standard in open-source video generation with Apache 2.0 licensing.

Initial explorations of Anthropic's new Computer Use capability
Anthropic unveils Claude 3.5 Sonnet and a new API for computer use, enhancing interaction with screenshots and addressing security risks.

Show HN: CyScout ā€“ Solidity Vulnerability Detection Powered by GitHub CodeQL
CoinFabrik's CyScout enhances CodeQL for Solidity, enabling effective vulnerability detection in smart contracts.

Show HN: LLM Deceptiveness and Gullibility Benchmark
A new benchmark evaluates LLMs' ability to generate and resist disinformation, testing major models like GPT-4 and Claude.

UK Ofcom: Social media algorithms must be adjusted to prevent misinformation fro
Ofcom calls for social media algorithm reforms to combat misinformation following riots linked to the Southport killings.

Marc Andreessen says AI model makers are in 'a race to the bottom'
Marc Andreessen warns that AI model makers are in a "race to the bottom," leading to commodification and poor business prospects.

Seeing Faces in Things: A Model and Dataset for Pareidolia
A new dataset of 5,000 pareidolic images reveals gaps in face detection between humans and machines, linking it to animal face recognition.

85% of Cursor AI in a Shell Script and Good Prompting
Integrate LLMs into software development with scripts for code context and file creation, emphasizing effective prompting techniques.

Artificial Kuramoto Oscillatory Neurons
Artificial Kuramoto Oscillatory Neurons enhance neural networks with dynamic synchronization, improving performance in various AI tasks.

ComfyUI Now Supports Stable Diffusion 3.5
ComfyUI now supports Stable Diffusion 3.5, featuring advanced models for enhanced image quality and user-friendly workflows.

Thom Yorke and Julianne Moore join creatives in AI warning
Thom Yorke and Julianne Moore join 10,500 creatives warning against AI's unlicensed use of their work, threatening artists' livelihoods.

Structured generation with Outlines, now in Rust
Outlines-core 0.1.0 launches structured generation in Rust and Python, enhancing speed, safety, and portability for LLM workflows.

Read more