🥟 Chao-Down #252 Meta to build open-source artificial general intelligence, OpenAI announces its first partnership with a university, Bing hasn't shown much market share increase despite AI features

Will independent verification of AI help address copyright concerns? A new non-profit founded by a former Stability AI VP aims to do just that.

Fairly Trained, a recently launched non-profit, is now offering a certification label for AI tools that are copyright compliant. Firms that disclose that they license the data used to train their AI models can submit their models to Fairly Trained for review and quality for certification. Those data sources can come from public-domain content or copyrighted data that companies have obtained permission to use.

The startup was founded by former Stability AI audio VP Ed Newton-Rex who left Stability because of disagreements with the startup’s stance on “fair use”. With the ongoing legal challenges from OpenAI, Microsoft, Google and others, some companies prefer to use AI models that they know have been solely trained on licensed or public-domain data.

So far, Fairly Trained has certified nine generative AI startups. Eight of which are focused on sound creation, while one specializes in image generation. So far, no firms specialize in text generation.

Will we start to see more of these independent certifications emerge, especially in the face of AI copyright concerns?

-Alex, your resident Chaos Coordinator.

What happened in AI? 📰

Meta’s new goal is to build artificial general intelligence (The Verge)

OpenAI announces first partnership with a university (CNBC)

ChatGPT's Hunger for Energy Could Trigger a GPU Revolution (WIRED)

Bing Search shows few, if any, signs of market share increase from AI features (Ars Technica)

Is A.I. the Death of I.P.? (The New Yorker)

Figure's humanoid robots are about to enter the workforce at BMW (newatlas.com)

Always be Learnin’ 📕 📖

A Product Marketing Masterclass with Superhuman's Alex Rodrigues (Substack)

Understanding and Coding Self-Attention, Multi-Head Attention, Cross-Attention, and Causal-Attention in LLMs (sebastianraschka.com)

12 techniques to reduce your LLM API bill and launch blazingly fast products (aitidbits.ai)

Projects to Keep an Eye On 🛠

cbh123/narrator: David Attenborough narrates your life (Github)

srikanth235/privy: Your private coding assistant (Github)

marimo-team/marimo: A reactive notebook for Python — run reproducible experiments, execute as a script, deploy as an app, and version with git. (Github)

The Latest in AI Research 💡

TrustLLM: Trustworthiness in Large Language Models (arxiv)

MoE-Mamba: Efficient Selective State Space Models with Mixture of Experts (arxiv)

Blending Is All You Need: Cheaper, Better Alternative to Trillion-Parameters LLM (arxiv)

The World Outside of AI 🌎

Workers are filming their layoffs, then posting them to TikTok. What could go wrong? (BBC)

The Billionaires Spending a Fortune to Lure Scientists Away From Universities (The New York Times)

Feds to Highway Signs: You Have Two Years to Stop Being Funny (WSJ)

China Unveils Plan for Massive ‘Silver Economy’ to Serve Elderly (Bloomberg)

How US minimum wage laws discriminate against young workers (qz.com)

Why Some People Keep Serious Illnesses Private (The New York Times)

One Last Bite 😋