Today's AI developments of note

Happy Wednesday everyone.

Have you tried asking your favorite AI chatbot “What’s a mythological parallel to this idea?” That was my latest little experiment last night, and Claude did a great job connecting supply chain analysis insight capture (random topic for fun) with the Three Fates in Greek Mythology. I learned some fun things and thought I’d share the prompt with you.

Here are today’s 6 stories that came to the top of our weighted analysis of AI community engagement. Claude thought the last one was reminiscent of the Labors of Hercules!

Today’s stories:

Weights & Biases, which counts OpenAI as a customer, lands $50M

"One of the more prolific AI and machine learning development platforms, Weights & Biases has secured a new tranche of cash from ex-GitHub CEO Nat Friedman and former Y Combinator partner Daniel Gross. Friedman and Gross, alongside existing investors Coatue, Insight Partners, Felicis, Bond, BloombergBeta and Sapphire, have invested $50 million in Weights & Biases in a strategic round that values the company at $1.25 billion. Bringing the startup’s total raised to $250 million, the investment comes as Weights & Biases prepares to launch Prompts, a new product designed to help users monitor and evaluate the performance of large language models." [TechCrunch]

Nvidia reveals new A.I. chip, says costs of running LLMs will 'drop significantly'

"Currently, Nvidia dominates the market for AI chips with over 80% market share, according to some estimates... But Nvidia’s chips are in short supply as tech giants, cloud providers and startups vie for GPU capacity to develop their own AI models....The new chip will be available from Nvidia’s distributors in the second quarter of next year, Huang said, and should be available for sampling by the end of the year. Nvidia representatives declined to give a price." [CNBC]

Meet the hackers who are trying to make AI go rogue

"In a windowless conference room at Howard University, AI chatbots were going haywire left and right. One exposed someone’s private medical information. One coughed up instructions for how to rob a bank. One speculated that a job candidate named Juan would have weaker 'interpersonal skills' than another named Ben. And one concocted an elaborate recounting of the night in July 2016 when it claimed Justin Bieber killed Selena Gomez." [WashingtonPost]

How AI is helping airlines mitigate the climate impact of contrails

Google's climate team, with American Airlines and Breakthrough Energy, has created a contrail reduction method. Tested live, it showed a notable contrail decrease, potentially cutting aviation's global warming effect by 20%. Despite using 2% more fuel, it could be a cost-effective warming solution. [Google.com]

Hugging Face Amplifies Reinforcement Learning Democratization Efforts

Hugging Face is boosting its dedication to democratizing Reinforcement Learning (RL) and RLHF (Reinforcement Learning from Human Feedback) with a new feature, Transformer Reinforcement Learning (TRL). RLHF is a powerful capability for improving AI performance and has been credited with much of ChatGPT's success, but the method is usually too expensive and complex for small organizations to perform. TRL from Hugging Face is a full stack library of tools to train transformer language models, from the Supervised Fine-tuning step (SFT), Reward Modeling step (RM) to the Proximal Policy Optimization (PPO) step. The library is integrated with Hugging Face transformers. [GitHub]

Kaggle Initiates $50,000 "LLM Science Exam" Challenge

The goal is to assess LLMs' self-testing abilities and usefulness in limited-resource environments. The challenge's dataset was formed from scientific Wikipedia snippets given to GPT3.5. Kaggle, Google's data science contest platform, shared a collection of resources developed by its community to learn more about LLMs in order to participate in this challenge. [Kaggle - LLM Science Exam] I can see how that’s a little like Hercules capturing the shaggy and wild Erymanthian Boar!

That’s it!

Thanks for reading,

Marshall