🥟 Chao-Down #36 AI scams on the rise, Museum of AI misalignment launches in SF, Google develops Universal Speech Model

More deepfake scams are coming to AI.

As AI gets more advanced, especially in the areas of speech and audio, bad actors are finding ways to profit off impersonating loved ones and duping them into giving up their money.

A couple in Canada reportedly lost $21,000 from an AI-generated voice that pretended to be their son. It’s causing real financial harm as it becomes almost impossible to tell whether a voice is real or not.

For the sake of our loved ones, we need to be careful.

-Alex, your resident Chaos Coordinator.

What happened in AI? 📰

Welcome to the Museum of the Future AI Apocalypse (WIRED)

Google’s one step closer to building its 1,000-language AI model (The Verge)

Microsoft to detail the ‘future of work with AI’ during March 16th event - (The Verge)

Microsoft Edge’s AI upscaling feature can improve low-resolution video - (The Verge)

The inside story of how ChatGPT was built from the people who made it | (MIT Technology Review)

Always be Learnin’ 📕 📖

Making model initialization faster than transferring from CPU (link)

200-Year-Old Math Opens Up AI’s Mysterious Black Box (IEEE)

Projects to Keep an Eye On 🛠

Datasets at your fingertips in Google Search – Google AI Blog (Google Blog)

Shark-NLP/OpenICL: OpenICL is an open-source framework to facilitate research, development, and prototyping of in-context learning. (Github)

The Latest in AI Research 💡

In-Context Instruction Learning (arxiv)

StyO: Stylize Your Face in Only One-Shot (arxiv)

Grounded Decoding - Guiding Text Generation with Grounded Models for Robot Control (grounded-decoding.github.io)

The World Outside of AI 🌎

Delta’s free in-flight wifi was a multiyear journey (Emerging Tech Brew)

Japan’s Push to Boost Birth Rates Is Falling Short (Bloomberg)

Amazon NFTs Will Be Tied to Real-world Assets, Token Possible (Blockworks)

One Last Bite 😋