Welcome (back) to AI Minds, a newsletter about the brainy and sometimes zany world of AI, brought to you by the Deepgram editorial team.

In this edition:

• 🎥 Veritasium: The Most Useful Thing AI Has Ever Done

• 🔊 Deepgram releases Nova-3: Setting a New Standard for AI-Driven Speech-to-Text

• 🌪️ Is it even possible for AI to predict the weather?

• 💻 Rapid Mobile App Development for Generative AI Agents on MIT App Inventor

• 🧠 Social Media Buzz: Elon Musk released Grok-3 just hours ago

• ⛏️ Twitter: An fully AI Society thrives in Minecraft and how AI approaches the economy

• 📲 Three new, trending AI apps for you!

• 🎙️ AI Minds Podcast with Abdulrahman Jamjoom, Co-Founder and CEO at Arini AI

Thanks for letting us crash your inbox; let’s party. 🎉

🎥 Veritasium: The Most Useful Thing AI Has Ever Done

AI is meant to solve problems. But what problem is the most important? Well, the YouTuber Veritasium has interviewed prominent researchers at Google DeepMind to come to this conclusion. Check it out!

⚡ Deepgram releases Nova-3, Extending Leadership in Voice AI for Enterprise Use Cases

Introducing Nova-3: Setting a New Standard for AI-Driven Speech-to-Text

Nova-3 advances Deepgram's industry-leading accuracy, extending its capabilities to a broader range of real-world enterprise use cases and challenging audio conditions. It is the first voice AI model to offer real-time multilingual transcription.

Further, this latest model is also the first to provide users with demonstrably effective and highly accurate self-serve customization—enabling instant vocabulary adaptation without model retraining.

Nova-3 delivers industry-leading performance with a 54.3% reduction in word error rate (WER) for streaming and 47.4% for batch processing compared to competitors. It was also preferred over Whisper on 7 out of 7 languages tested—reaching as high as an 8-to-1 preference on certain languages.

To try Nova-3, click here!

🔍  AI Predicting the Weather and MIT’s App Inventor

AI Foundation Models for Weather and Climate: Applications, Design, and Implementation - Despite the progress of AI in the past few years, we are still in the nascent stages of a generalizable AI model for global Earth system models, regional climate models, and mesoscale weather models. This paper reviews current state-of-the-art AI approaches, primarily from transformer and operator learning literature in the context of meteorology.

Rapid Mobile App Development for Generative AI Agents on MIT App Inventor - The authors of this paper present a methodology for the rapid development of AI agent applications using the development platform provided by MIT App Inventor.

🐝 Social Media Buzz: Elon Releases Grok 3 and an AI-ruled Minecraft Society

Elon Musk releases Grok 3 tonight, a few hours after this newsletter is set to be published. Reviews on this technology should be coming next week! What are your expectations for Grok 3?

📲 Three new, trending AI Apps for you!

PimEyes provides a powerful AI-driven solution, allowing individuals to track and monitor their images online. It functions as an advanced facial recognition search engine, scanning the web for images that match a given face.

Emvoice is a vocal synthesizer plugin and app that allows you to compose vocal music just by entering lyrics and notes. The vocals are generated using AI and performed by Emvoice's virtual singers such as Keela, Lucy, Jay, and Thomas.

Rezi is an innovative, AI-powered resume builder that simplifies the process of creating professional, ATS-optimized resumes. Designed to help job seekers stand out, Rezi automates everything from resume writing to keyword optimization, enabling users to craft job-specific resumes in just minutes.

🎤 The AI Minds Podcast

Abdulrahman Jamjoom, Co-Founder and CEO at Arini AI, shares his journey from Saudi Arabia to Jordan and then to the U.S. He recounts his experiences at Threads and his determination to create a startup before even graduating from Harvard, which eventually led him to co-found Arini AI.

Abdul talks about the technical challenges of building scalable AI systems, highlighting the complexities involved in managing interruptions, minimizing latency, and ensuring accurate transcription, particularly in sensitive contexts such as insurance data.