AIR#178 - ChatGPT's Code Wizardry & A $400M Medicaid Scandal 🌟
Hey there!
Here's the latest AI news for today. Enjoy!
Today's top stories
🔥 OpenAI is good at unminifying code
ChatGPT effectively reverse-engineered minified JavaScript, generating a readable TypeScript version of the code.
🔥 Judge rules $400M algorithmic system illegally denied Medicaid benefits
A judge ruled that Tennessee's $400M algorithmic Medicaid system unlawfully denied benefits to thousands due to errors.
🔥 Dawn of a new era in Search: Balancing innovation, competition, and public good
A judge ruled Google a monopoly, sparking debate on search competition and proposing access to its Search Index for innovation.
🔥 Update on Llama adoption
Llama's adoption surged 10x in 2024, nearing 350 million downloads, solidifying its position as a leading open-source AI model.
🔥 Show HN: turn videos into ASCII art (open source, js+canvas)
Transform videos into ASCII art with this open-source tool, allowing uploads or webcam use, customizable settings, and exports.
🔥 100M Token Context Windows
Magic's ultra-long context model, LTM-2-mini, processes 100M tokens, enhancing AI's code synthesis and efficiency significantly.
🔥 Major Sites Are Saying No to Apple's AI Scraping
Major news outlets are opting out of Apple's AI scraping, citing concerns over data usage and copyright protection.
Show HN: LLM-Term – Simple Rust-based CLI assist tool
LLM-Term is a Rust CLI tool that generates and executes terminal commands using OpenAI's language models.
Applying to 1k Jobs with AI
Santiago warns against applying to 1,000 jobs with AI, claiming it leads to being blacklisted and reveals weak candidacy.
Show HN: MinutesLink – AI note taker for online calls
MinutesLink is an AI note taker for online calls, providing accurate transcripts and summaries to enhance meeting efficiency.
MiniTorch – a DIY teaching library for machine learning engineers
MiniTorch is a DIY Python library for learning deep learning concepts, featuring assignments and visualization tools.
Show HN: KTransformers–236B Model and 1M Context LLM Inference on Local Machines
KTransformers offers a flexible framework for local LLM inference, supporting 1M context with optimized performance on desktops.
Show HN: A directory to find open source AI projects easier
A new directory, AIExh, simplifies discovering open-source AI projects, featuring over 2,726 innovative resources.
Show HN: Geppetto v0.2.5 Claude, ChatGPT, and Gemini OSS Companion in Your Slack
Geppetto v0.2.5 integrates LLMs like ChatGPT and Claude into Slack, enhancing productivity and collaboration for users.
Midjourney says it's 'getting into hardware'
Midjourney announces plans to enter hardware development, hinting at potential projects related to AI and mixed reality.
Generative AI coding startup Magic lands investment from Eric Schmidt
Generative AI startup Magic secures $320M investment from Eric Schmidt and others to enhance coding automation and supercomputing.
LTM-2-Mini is our first model with a 100M token context window
LTM-2-Mini features a groundbreaking 100M token context window, equivalent to 10M lines of code or 750 novels.
TypingMind introduces Artifacts for any AI model, not just Claude
TypingMind now offers Artifacts for any AI model, expanding functionality beyond just Claude.
Apple in talks to invest in OpenAI as ChatGPT comes to iOS 18
Apple is negotiating to invest in OpenAI as ChatGPT integrates with iOS 18, enhancing its AI capabilities.
How to become a great LLM engineer
To excel as an LLM engineer, cultivate curiosity, resilience, and flexibility while leveraging AI tools and debunking common myths.
OpenAI and Anthropic sign safety agreement with US Government
OpenAI and Anthropic sign safety agreements with the U.S. government to enhance AI safety research and evaluation.
Show HN: Let me plex that for you – Teach friends and others to use Perplexity
A new tool, "Let Me Plex That," helps users teach friends how to use Perplexity, coined by founder Greg Kogan.
AI generates covertly racist decisions about people based on their dialect
AI models exhibit covert racism by making biased decisions based on dialect, particularly against African American English speakers.
Silicon Valley Confronts a Grim New A.I. Metric
Silicon Valley grapples with "p(doom)," a new metric assessing the likelihood of AI-induced catastrophe.
OpenAI in talks to raise funding that would value it at more than $100B
OpenAI seeks funding to surpass a $100B valuation, led by Thrive Capital's $1B investment, amid plans for a new search engine.
New AI model can hallucinate a game of 1993's Doom in real time
Researchers unveil GameNGen, an AI model that can real-time simulate 1993's Doom, revolutionizing video game graphics synthesis.
Kan 2.0: Kolmogorov-Arnold Networks Meet Science
KAN 2.0 proposes a framework merging Kolmogorov-Arnold Networks with scientific discovery, enhancing feature identification and formula extraction.
Show HN: I built a tool that creates and edits videos using text prompts
Cliptalk enables users to create and edit videos from text prompts quickly, automating the process with AI tools.
Introduction to Mechanistic Interpretability
Mechanistic interpretability aims to unveil neural networks' internal processes, enhancing AI safety and decision-making transparency.
Nvidia Discusses Joining OpenAI's Latest Funding Round
Nvidia is in talks to join a funding round for OpenAI, potentially valuing the startup over $100 billion, alongside Apple and Microsoft.
3rd circuit rules companies are liable for their algorithms
A U.S. court revives a lawsuit against TikTok, holding it liable for algorithmically promoting a dangerous challenge linked to a child's death.
BiomarkerAI: Turn blood test pdfs into insights and a dashboard visualization
SiPhox Health's BiomarkerAI simplifies blood test results into actionable insights, enhancing health understanding and tracking.