🥟 Chao-Down #105 OpenAI CEO calls for collaboration with China, AI giants give UK early access to foundation models, Researchers warn of 'model collapse' as AI trains on AI-generated content

Is the future of education with Khanmigo?

Khan Academy's experimental chatbot tutor, Khanmigo, recently launched with sixth graders at Khan Lab School in Palo Alto, California.

These chatbots aim to simulate one-on-one human tutoring, using artificial intelligence to provide customized academic support. The bots can respond to students in clear, smooth sentences and are designed specifically for school use.

Proponents believe these automated tutoring systems have the potential to revolutionize classroom teaching and learning, allowing self-directed students to enhance their skills and explore new subjects at their own pace.

However, critics argue that these unproven systems may lead to errors, cheating, and diminished teacher roles, turning students into test subjects for algorithmic education experiments.

Will AI tutors like Khanmigo be able to fill the gap by providing students an ever-evolving personalized tutor? Or are they still missing some fundamental element that you can only find in person-to-person instruction?

-Alex, your resident Chaos Coordinator.

What happened in AI? 📰

OpenAI, DeepMind and Anthropic to give UK early access to foundational models for AI safety research (TechCrunch)

The AI feedback loop: Researchers warn of 'model collapse' as AI trains on AI-generated content (VentureBeat)

A Tale of Two Economies - For anyone working on AI, it’s the best of times, but for others, it’s the worst of times. (WSJ)

OpenAI CEO Calls for Collaboration With China to Counter AI Risks (WSJ)

Salesforce (CRM) Touts AI Strategy, Doubles Generative AI Venture Fund (Bloomberg)

Always be Learnin’ 📕 📖

What is a Vector Database? (Pinecone)

Understanding DeepMind's Sorting Algorithm (justine.lol)

Test Your Startup Idea With AI: Must-Know Tips and Tools (Entrepreneur's Handbook)

Why trying to "shape" AI innovation to protect workers is a bad idea (noahpinion.blog)

Projects to Keep an Eye On 🛠

AI4Finance-Foundation/FinGPT: Data-Centric FinGPT. Open-source for open finance! Revolutionize 🔥 We're just getting started! We'll soon be releasing the trained model. (Github)

go-skynet/LocalAI: :robot: Self-hosted, community-driven, local OpenAI-compatible API. Drop-in replacement for OpenAI running LLMs on consumer-grade hardware. (Github)

Bard updates: More accurate responses, export to Google Sheets (blog.google)

The Latest in AI Research 💡

Surfacing Biases in Large Language Models using Contrastive Input Decoding (arxiv)

Can foundation models label data like humans? (huggingface.co)

RL4F: Generating Natural Language Feedback with Reinforcement Learning for Repairing Model Outputs (arxiv)

The World Outside of AI 🌎

Sol Reader is a VR headset exclusively for reading books (TechCrunch)

Why We Tell Lies More on a Laptop Than on a Cellphone (WSJ)

Tip screen pressure is driving Americans to leave smaller tips (axios)

How Gas Stoves Became Part of America’s Culture Wars - (Bloomberg)

Want a career saving the planet? Become an electrician. (The Washington Post)

One Last Bite 😋