Good morning. It’s Monday, June 10th.

Did you know: On this day in 1985, Microsoft and IBM entered into a joint development agreement to co-develop the OS/2 operating system. This agreement fell through in the early '90s, after which IBM became the sole developer of OS/2.

In today’s email:

• Apple Intelligence

• Open-Source Chore Robot

• OpenAI “Thinking” Map

• 5 New AI Tools

• Latest AI Research Papers

You read. We listen. Let us know what you think by replying to this email.

Apple's WWDC is reportedly all about Apple Intelligence

Apple's WWDC is anticipated to spotlight extensive AI advancements dubbed "Apple Intelligence" for iPhone, iPad, and Mac, per Bloomberg's Mark Gurman. The keynote is rumored to heavily emphasize AI, demonstrating functionalities designed to optimize user experience.

Key areas of interest include improvements to Siri, privacy considerations, and on-device AI capabilities. Apple's approach, focusing on efficiency and on-device processing, stands in contrast to competitors relying on large server farms.

The conference is also expected to address other software upgrades and potentially expand on the Vision Pro virtual and augmented reality headset. These advancements aim to leverage AI software to invigorate hardware sales – with a focus on user convenience, privacy, and device compatibility – in a softening market.

However, expectations for dedicated AI functionalities like image/video generation or new hardware announcements at WWDC remain modest. Read more.

Hugging Face and Pollen Robotics Show Off Open Source Robot That Does Chores  

Hugging Face and Pollen Robotics join forces to reveal their collaborative masterpiece: an open-source humanoid robot named Reachy2. Developed under Hugging Face's Le Robot initiative, Reachy2 showcases the symbiotic blend of AI and robotics. Remi Cadene, a former Tesla engineer, elucidated the innovative training regimen behind Reachy2, combining teleoperation with machine learning algorithms.

This amalgamation enables Reachy2 to execute household chores and safely interact with humans, underpinned by 40,000 to 60,000 steps of rigorous training. Leveraging open-source ethos, Hugging Face and Pollen Robotics democratize robotics by releasing datasets and trained models for community use.

With Reachy2 poised for imminent release, the collaboration underscores Pollen Robotics' commitment to accessible and ethical robotics. Read more.

OpenAI's New Method Shows How GPT-4 "thinks" In Human-understandable Concepts

OpenAI introduces a method to dissect the internal mechanisms of GPT-4, revealing 16 million potentially interpretable patterns, aiding in understanding AI model safety and robustness. Employing scaled "sparse autoencoders," OpenAI aims to transform complex GPT-4 activation patterns into concise features understandable by humans. Each feature ideally corresponds to a human-interpretable concept within GPT-4.

While this method identifies specific concepts like human flaws and ML training logs, complete model behavior mapping remains challenging. The key obstacle lies in scaling the technique to billions or trillions of features. OpenAI releases a paper, Github source code, and an interactive visualizer for the autoencoder's learned features. Competitor Anthropic also explores model interpretability, emphasizing scalability challenges. Read more.

Etcetera: Stories you may have missed

Princeton’s AI Unlocks New Levels of Performance in Fusion Reactors

Alibaba's Qwen2 sets new standards for open source language models

Harvard, MIT, and Wharton research reveals pitfalls of relying on junior staff for AI training  

Meta to use Instagram and Facebook posts from as far back as 2007 to train artificial intelligence tools

This is AI's 'next wave,' according to Nvidia CEO Jensen Huang

Google Deepmind says open-ended AI is key to achieving superintelligence

OpenAI seems awfully defensive about its AI voice engine

Microsoft's Recall feature will now be opt-in and double encrypted after privacy outcry

DuckDuckGo will now allow you to anonymously use ChatGPT, Claude and Meta AI for free. Here's how it works

Peek is an AI browser tab organizer, auto-grouping and summarizing tabs, integrating notes, prioritizing tasks, and enhancing productivity in major browsers.

FANCY provides subscription-based design services for Web3, AI, Fintech, and eCommerce, offering expert designers, fast communication, and unlimited revisions.

DomoAI is an AI-powered art generation tool, transforms text to images, static images to videos, and anime to realistic photos, enhancing creativity and saving design time.

LocalhostAI is a free, private, offline AI assistant enhancing productivity with Chrome and Gemini Nano, offering fast, open-source AI models.

OpinioAI is an AI-powered research platform offering data analysis, creative evaluation, and insight generation. Key features include real-time data processing, predictive analytics, customer sentiment analysis, idea testing, and integration capabilities.

📄 Large Language Model Confidence Estimation via Black-Box Access

📄 WildBench: Benchmarking LLMs with Challenging Tasks from Real Users in the Wild ​​

📄 Proofread: Fixes All Errors with One Tap

📄 GenAI Arena: An Open Evaluation Platform for Generative Models

📄 CRAG -- Comprehensive RAG Benchmark

Thank you for reading today’s edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, apply here.