🥟 Chao-Down #149 Researchers build AI to combat California wildfires, Zoom updates terms of service to say no customer data is used for training AI, Meta disbands protein-folding team

It looks like Zoom’s learned the error of its ways in their terms of service.

Originally, Zoom’s change to their TOS appeared to give the company royalty-free rights in perpetuity for customer video calls and presentations for the purposes of training large AI models.

Now in a more definitive statement, Zoom says that it “does not use any of your audio, video, chat, screen-sharing, attachments or other communications like customer content (such as poll results, whiteboard, and reactions) to train Zoom’s or third-party artificial intelligence models.”

What’s true is that building these large foundation models is a data-hungry endeavor, but companies like Zoom will need to find different ways to gather such data or face customer backlash.

-Alex, your resident Chaos Coordinator.

What happened in AI? 📰

California turns to AI to help spot wildfires (Reuters)

Amazon is racing to catch up in generative A.I. with custom AWS chips (CNBC)

Zoom AI Backlash: TOS Update Says No Customer Content Used in Training (Variety)

ChatGPT expands its ‘custom instructions’ feature to free users (TechCrunch)

Meta disbands protein-folding team in shift towards commercial AI (Financial Times)

The New York Times forbids using its content to train AI models (The Verge)

People Are Having Sex in Robotaxis. Nobody Is Talking About It (sfstandard.com)

Always be Learnin’ 📕 📖

Why Do People Think the Economy is Bad? (Substack)

End-to-End Secure Evaluation of Code Generation Models (MosaicML)

The Race to Capture Value: Cloud Lessons for the AI Era (Andreessen Horowitz)

Projects to Keep an Eye On 🛠

simonmesmith/agentflow: Complex LLM Workflows from Simple JSON. (Github)

joonspk-research/generative_agents: Generative Agents: Interactive Simulacra of Human Behavior (Github)

chain-ml/council: Council is an open-source platform for the rapid development and robust deployment of customized generative AI applications (Github)

The Latest in AI Research 💡

Full Stack Optimization of Transformer Inference: a Survey (arxiv.org)

In-Context Alignment: Chat with Vanilla Language Models Before Fine-Tuning (arxiv)

Your Negative May not Be True Negative: Boosting Image-Text Matching with False Negative Elimination (arxiv)

The World Outside of AI 🌎

'Blind Side' subject Oher alleges Tuohys made millions off lie (ESPN)

As Maui Tries to Recover From Wildfires, Residents Say Aid Is Scant (The New York Times)

Why Being an Airbnb Host Is Much Harder Than in the Past (Bloomberg)

Women’s Problem Drinking Is Catching Up to Men’s (WSJ)

A look inside the world of rare trading cards and the biggest buy this year (NPR)

Lyft doesn't mind its drivers earning less and less (qz.com)

One Last Bite 😋