🥟 Chao-Down #33 OpenAI launches ChatGPT and Whisper API, OpenAI will no longer train on customer data

Today is all about OpenAI!

While the launch of ChatGPT and the speech-to-text model Whisper is getting all the attention, one other announcement that is an eye-opener is OpenAI making a public statement that they will no longer use customer data to train their models by default.

Specifically, any data that is sent to OpenAI’s various model APIs won’t be used for “service improvements” unless a company or individual opts-in. While they say that these aren’t necessarily changes, it shows the sensitivity that people have towards pushing their personal data to public endpoints.

It’s good that OpenAI is spelling this out now, but it probably should have been done right from the get-go.

-Alex, your resident Chaos Coordinator.

What happened in AI? 📰

Addressing criticism, OpenAI will no longer use customer data to train its models by default (TechCrunch)

OpenAI debuts Whisper API for speech-to-text transcription and translation | (TechCrunch)

OpenAI launches an API for ChatGPT, plus dedicated capacity for enterprise customers (TechCrunch)

YouTube’s new leader teases AI tools that can virtually swap outfits (The Verge)

Nvidia’s latest GPU drivers can upscale old blurry YouTube videos (The Verge)

Always be Learnin’ 📕 📖

Prompt Vibes - Discover, Learn, and Test Different ChatGPT Prompts (link)

Projects to Keep an Eye On 🛠

PromptLoop | AI in Google Sheets™ and Excel™ with a single formula (link)

Detoxifying a Language Model (HuggingFace Spaces)

World Building With GPT (link)

The Latest in AI Research 💡

Collage Diffusion: Fine-Grained Control Over Diffusion-Based Image Generation (arxiv)

Teaching old labels new tricks in heterogeneous graphs – (Google AI Blog)

The World Outside of AI 🌎

How loneliness reshapes the brain (Quanta Magazine)

The KFC Double Down is coming back after a 10-year hiatus | (CNN Business)

One Last Bite 😋