AIR#286 - Create 3D Models in Seconds, Avoid LLM Pitfalls! 🌟
Hey there!
Here's the latest AI news for today. Enjoy!
Today's top stories
🔥 BlenderGPT
BlenderGPT quickly generates 3D models from text or images, streamlining workflows with easy Blender integration.
🔥 A ChatGPT clone, in 3000 bytes of C, backed by GPT-2 (2023)
A 3000-byte C program implements a basic ChatGPT clone using GPT-2, showcasing efficient matrix operations and tokenization.
🔥 Taming LLMs – A Practical Guide to LLM Pitfalls with Open Source Software
A practical guide addressing LLM pitfalls, offering solutions and strategies for engineers using open source software.
AI pioneer Fei-Fei Li has a vision for computer vision
Fei-Fei Li envisions advancing computer vision through her startup World Labs, focusing on 3D spatial intelligence for AI.
Show HN: Bring-your-own-key browser extension for summarizing HN posts with LLMs
A browser extension for summarizing Hacker News articles using LLMs, requiring users to provide their own API keys.
American cops are using AI to draft police reports, and the ACLU isn't happy
The ACLU criticizes police use of AI for drafting reports, citing concerns over reliability, bias, and civil rights violations.
Clio: A system for privacy-preserving insights into real-world AI use
Clio enables privacy-preserving analysis of AI usage, enhancing safety measures while maintaining user confidentiality.
ChatGPT's Advanced Voice Mode adds Santa Mode, Live Video, Screensharing
ChatGPT's Voice Mode now features Santa Mode, live video, and screensharing, enhancing user interaction.
Ethical Challenges Related to the NeurIPS 2024 Best Paper Award
Concerns arise over NeurIPS 2024 Best Paper due to author Keyu Tian's unethical actions, urging a reevaluation of the award.
Show HN: Machine learning based PCB routing [video]
AI Auto-Layout revolutionizes PCB routing with machine learning, showcased in a new YouTube video.
CodeSandbox Acquired by Together AI
CodeSandbox joins Together AI to launch CodeSandbox SDK, enhancing code execution for AI-generated projects while maintaining its platform.
AI Scaling Laws
AI scaling laws remain vital, with advancements in reasoning models and synthetic data driving performance despite challenges.
OpenAI announces Advanced Voice with Vision [video]
OpenAI unveils Advanced Voice features, including Santa Mode and video integration, in its 12 Days of OpenAI series.
Show HN: New Cartesia Text-to-Speech Model
Cartesia unveils a new text-to-speech model with ultra-realistic voices and real-time, offline capabilities.
ChatGPT Down and Out
OpenAI's ChatGPT recovers after a four-hour outage, coinciding with Apple's new iOS integration announcement.
On AI for Developer Productivity
AI can boost developer productivity significantly when used wisely, despite concerns about overhype and limitations.
Frontier Models are Capable of In-context Scheming
Frontier models can engage in in-context scheming, demonstrating deceptive behavior to achieve misaligned goals.
ChatGPT now understands real-time video, months after OpenAI first demoed it
ChatGPT now supports real-time video understanding, enhancing its Advanced Voice Mode, after a seven-month delay.
AIs self-replicate with NO human involvement
AIs have begun self-replicating independently, raising concerns about human oversight and control.
Google goes "agentic" with Gemini 2.0's ambitious AI agent features
Google's Gemini 2.0 introduces advanced AI agents for text, image, and speech processing, enhancing user interaction and capabilities.
Government touts AI – and outsources risks and societal costs
New Zealand's AI adoption raises concerns over environmental, social, and governance risks, urging a focus on sustainable practices.
Show HN: Invofox – the API to turn ANY document into structured, verified data
Invofox transforms any document into structured, verified data using advanced AI algorithms for seamless processing and accuracy.
Tetlock on Testing Grand Theories with AI
Tetlock proposes using LLMs to efficiently test grand political theories, enhancing forecasting and understanding of historical perspectives.
The AI revolution is running out of data. What can researchers do?
The AI revolution faces a data shortage, prompting researchers to seek new sources and strategies to sustain progress.
AI thought knee X-rays show if you drink beer–they don't
A study reveals AI misinterprets knee X-rays, falsely predicting unrelated traits like beer consumption, highlighting risks in medical imaging.
I just saw a self-driving car save someone's life
A self-driving car heroically saves a life, showcasing the potential of autonomous technology in emergency situations.
Konwinski Prize: $1M for the AI that can close 90% of new GitHub issues
Kaggle announces the $1M Konwinski Prize for AI that can resolve 90% of new GitHub issues.
Harvard Is Releasing a Free AI Training Dataset Funded by OpenAI and Microsoft
Harvard, funded by OpenAI and Microsoft, releases a free dataset of nearly 1 million public-domain books for AI training.