🥟 Chao-Down #80 Bing's AI chatbot teases multimodality, Google's push to make search more personal, Cedars-Sinai develops AI to detect heart attacks

In today’s edition of Chao-Downs, I feature many of the fully open-source large language models that have commercially friendly licenses and can be used by hobbyists to companies-alike to build upon and create custom AI with.

Last time, I covered the leaked memo from a Google researcher claiming that Google (and any other large tech company) has no moat in AI. A big reason for that is the rise of open-source. While the models coming from OSS aren’t the most performant nor the highest quality, they are effectively the mid-tier, budget smartphone equivalent to the powerful flagship versions that come from Google, OpenAI, Anthropic and others. If this was even one year ago, the models coming from open source would be considered state-of-the-art.

What’s true is that the speed at which open source is moving is incredible. And a well-coordinated collective community effort might very well dethrone the likes of a tech giant. Certainly, there are exciting times ahead.

-Alex, your resident Chaos Coordinator.

What happened in AI? 📰

Microsoft’s Bing chatbot gets smarter with restaurant bookings, image results, and more (The Verge)

Google Adding AI Chatbot, More Video to Stay On Top of Search (WSJ)

America Forgot About IBM Watson. Is ChatGPT Next? (The Atlantic)

OpenAI’s regulatory troubles are just beginning (The Verge)

OpenAI contractors make $15 to train ChatGPT (NBC News)

Can Artificial Intelligence Predict Heart Attack Risk? (cedars-sinai.org)

Always be Learnin’ 📕 📖

LoRA: Low-Rank Adaptation of Large Language Models (arxiv)

What is a Vector Database? | (Pinecone)

5 red flags that make recruiters scroll past your LinkedIn profile (Fast Company)

Projects to Keep an Eye On 🛠

eugeneyan/open-llms: 🤖 A list of open LLMs available for commercial use. (Github)

yxuansu/OpenAlpaca: OpenAlpaca: A Fully Open-Source Instruction-Following Model Based On OpenLLaMA (Github)

StarCoder: A State-of-the-Art LLM for Code (huggingface.co)

Introducing MPT-7B: A New Standard for Open-Source, Commercially Usable LLMs (MosaicML)

The Latest in AI Research 💡

Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizes (arxiv)

VPGTrans - Transfer Visual Prompt Generator across LLMs (link)

Search-in-the-Chain: Towards Accurate, Credible and Traceable Large Language Models for Knowledge-intensive Tasks (arxiv)

Distill or Annotate? Cost-Efficient Fine-Tuning of Compact Models (arxiv)

The World Outside of AI 🌎

The End of the Free-Returns Era (The Atlantic)

Sam Altman says remote work ‘experiment’ was a mistake (Fortune)

Covid global health emergency is over, WHO says - (BBC News)

Online Shopping Scams Aimed at Young People Are Growing—Here's How to Avoid the Scammers - (WSJ)

One Last Bite 😋