Good morning. It’s Friday, May 3rd.

Did you know: On this day in 2002, HP completed its buyout of Compaq Computer Corp. The transaction was valued at $18.7 billion.

In today’s email:

• Rabbit R1 - Just an Android App?

• Altman Interview

• Sora Music Video

• Open-Source Voice Generator

• 10 New AI Tools

• Latest AI Research Papers

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

Rabbit R1 AI Box Revealed To Just Be An Android App

The Rabbit R1, positioned as an AI device to supplant smartphones, has been scrutinized for its utilization of the Android Open Source Project (AOSP). Despite efforts by Rabbit Inc. to distance itself from conventional smartphone operating systems, analysis reveals the R1 essentially operates on a smartphone OS within a constrained $200 device.

While Rabbit Inc. seems displeased with this disclosure, the technical choice to employ AOSP aligns with industry norms for mobile hardware, leveraging its open-source nature and comprehensive feature set. By leveraging AOSP, the Rabbit R1 gains established functionalities such as mobile network connectivity, touchscreen management, and efficient system-on-chip utilization, without the complexities and restrictions associated with Google Play integration. Read more.

'GPT-4 Is The Dumbest Model Any Of You Will Ever Have To Use' Declares OpenAI CEO Sam Altman As He Bets Big On A Superintelligence

OpenAI's CEO Sam Altman describes ChatGPT as the "dumbest" it will ever be, signaling significant investment in AI's future.

Watch the full talk here [45min]:

In a Stanford seminar, Altman addresses concerns about rising AI development costs and emphasizes the importance of providing capable tools for users to innovate. He jokes about financial concerns within OpenAI, highlighting a focus on product quality over expenditure. Altman stresses iterative deployment and societal co-evolution with technology, suggesting a need for continuous improvement rather than waiting for perfect solutions. Read more.

The First Music Video Generated With OpenAI's Unreleased Sora Model Is Here

OpenAI's unreleased AI model, Sora, is showcased in a groundbreaking music video created by writer/director Paul Trillo for indie artist Washed Out. Trillo reveals that the video comprises 55 clips out of 700 generated by Sora, stitched together using Adobe Premiere.

While Adobe plans to integrate Sora into Premiere Pro, Trillo's workflow currently relies on third-party software for clip generation. Washed Out expresses excitement about incorporating Sora into future projects, emphasizing its potential to innovate storytelling. Read more. 

OpenVoice V2 Breaks The Language Barrier And Redefines Voice Interaction

Myshell-ai recently released OpenVoice V2, an updated version of its text-to-speech software. Building on the capabilities of V1, V2 delivers improved audio quality thanks to a revised training approach.

V2 now supports multiple languages natively, including English, Spanish, French, Chinese, Japanese, and Korean. Both V1 and V2 are now freely available for commercial use under the MIT License, effective April 2024.

OpenVoice V2 introduces several advanced functionalities, such as precise tone color replication, flexible voice style manipulation, and zero-shot cross-lingual voice cloning. These features empower developers and researchers to undertake tasks like generating emotionally nuanced speech and synthesizing speech across different languages seamlessly. Read more.

🖇️ Etcetera: Stories you may have missed

> Meshy-3 is Here: Ultra Realistic Sculptures, PBR, and New Image to 3D Pipeline

> Tesla's next-gen Dojo AI training tile is in production

> Anthropic's AI assistant Claude gets an iOS app and new team plan for businesses  

> Microsoft, Google and Meta Bet on Fake Data to Build AI Models

> Sam Altman says helpful agents are poised to become AI’s killer function

> Microsoft taps Sanctuary AI for general-purpose robot research

> China tops the U.S. on AI research in over half of the hottest fields

> Microsoft made the biggest renewable energy agreement ever to fuel its AI ambitions

> Ideogram Pro launches with 12K fast AI image generations monthly

> Microsoft’s OpenAI investment was triggered by Google fears, emails reveal

> Nvidia's free local chatbot adds new AI models, image search, and voice input

> Google urges US to update immigration rules to attract more AI talent

> Microsoft bans US police departments from using enterprise AI tool for facial recognition

> The insect farmers turning to AI to help lower costs

> StarCoder2-Instruct: Fully Transparent and Permissive Self-Alignment for Code Generation

> Tabnine and Atlassian reveal new generative AI tools for developers 

Count is AI-driven accounting software for small businesses, automating tasks like transaction categorization and offering real-time reporting. It integrates with 12,000+ financial institutions.

QueryPal is an AI chat assistant for Slack, leveraging company data from Google Drive, Notion, Jira, and Confluence. Customizable, secure, and launched in SaaS productivity.

AI Assistant and Bot Builder enables no-code creation of custom AI assistants, integrating with tools and databases, and supporting various platforms. Compatible with any website or app builder.

Nendo is an AI tool for creating sample packs from music libraries, automating stem extraction and offering exploration features. Integrates with DAWs and DJ apps.

1Flow facilitates targeted, in-product surveys with AI, enabling quick, actionable feedback. Loved by product managers for cutting research timelines and founders for better insights.

Playbook x GPT facilitates advanced media management by harnessing GPT's AI for intuitive search, retrieval, and organization of images and videos.

websim.ai offers a comprehensive platform for URL and content manipulation. Renowned for its versatility, it enables users to handle diverse inputs effectively, fostering both practical and artistic applications.

Monterey AI 2.0 is an advanced tool for aggregating, triaging, and analyzing user feedback from multiple sources, providing real-time, deep insights.

Mockey AI offers 5000+ AI-generated mockup templates across 25+ categories. Features include premium bundles, editable backgrounds, high-quality downloads, and ad-free experience.

WaxWing serves as an AI copilot for marketing tasks, leveraging contextual understanding to provide personalized strategies, plans, and execution support.

📄 LLM-AD: Large Language Model based Audio Description System  

📄 Prometheus 2: An Open Source Language Model Specialized in Evaluating Other Language Models ​​

📄 NeMo-Aligner: Scalable Toolkit for Efficient Model Alignment 

📄 FLAME: Factuality-Aware Alignment for Large Language Models

📄 Customizing Text-to-Image Models with a Single Image Pair

📄 Data-Efficient Multimodal Fusion on a Single GPU (Layer 6 AI)

📄 Automatic Creative Selection with Cross-Modal Matching (Apple)

📄 STT: Stateful Tracking with Transformers for Autonomous Driving (Google)  

📄 Self-Play Preference Optimization for Language Model Alignment

📄 Spectrally Pruned Gaussian Fields with Neural Compensation

AI Creates Comics

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.