• Chaos Theory
  • Posts
  • 🥟 Chao-Down #6: Scale AI hosts hackathon for building with LLMs, A16z asks "Who Owns the Generative AI Platform?"

🥟 Chao-Down #6: Scale AI hosts hackathon for building with LLMs, A16z asks "Who Owns the Generative AI Platform?"

Plus, Google calls on its founders for AI fight.

January 23, 2023

In what many are calling a “SF is back” moment, over the weekend, Scale AI hosted an AI hackathon focused on creating projects with large language models. With a huge turnout,

Image

Some of the most interested projects included:

  • Rap Battle - Pick any two people and generate a rap battle on the fly. (link)

  • GPT-3 Auditor - Scanning code for vulnerabilities with LLMs. (link)

  • Pierre Bhat, a prolific AI coder who roams GitHub resolving issues with PRs (link)

  • ActGPT - Chatbot that can perform tasks on browser. (link)

  • Adding long-term memory of relevant entities to make chatbots smarter (link)

  • And many more interesting use cases. (link)

It’s great to see the pure energy and excitement of gathering builders together ands seeing what they can create in a short period of time. Hopefully we’ll be seeing more in the times to come.

-Alex, your resident Chaos Coordinator

What happened in AI? 📰

Google calls in help from founders for AI fight (NYTimes)

Who Owns the Generative AI Platform? (A16z)

Rocket League devs say they’re cracking down on AI bots (Polygon)

Always be Learnin’ 📕 📖

  • A playbook for systematically maximizing the performance of deep learning models. (Github)

  • Tutorial - Theory and Practice of Efficient and Accurate Dataset Construction (NeurIPS)

  • Transformer models: an introduction and catalog — 2023 Edition - AI, software, tech, and people, not in that order… by X (amatriain.net)

Projects to Keep an Eye On 🛠

StockGPT, an AI financial research tool trained on all of Tesla’s earnings reports (link)

AI can read the codebase instead and tell you exactly what to change where (Github)

Feat 1

The Latest In AI Research 💡

  • Multiview compressive coding for 3D reconstruction. (link)

  • RecolorNeRF: Layer decomposed radiance field for efficient colour editing of 3D scenes. (link)

  • AtMan: Understanding transformer predictions through memory efficient attention manipulation. (link)

  • Self-supervised learning from images with a joint embedding predictive architecture. (link)

The World Outside of AI 🌎

Google officially shuts down Stadia, its cloud gaming service (link)

Rebalancing Chip Supply Chains will take decades, Intel CEO says (WSJ)

One Last Bite 😋