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:

• 📚 GPT-4 used to suck at math, but now…

• 🧠 Building “Living Software Systems” with Generative and Agentic AI

• 🤖 Multimodal Agent AI: Should we be worried or hyped?

• 🔊 Upcoming Webinar! Voice AI Labs and Conversational AI

• 🎙️ The Best (and Worst!) Text-to-Speech APIs

• 🎹 Tutorial: The Art and Magic of Finetuning LLMs

• 🐦 Social Media Buzz: Agentic AI and New Robots

• 📲 Three New Trending AI Apps for You!

• 🎤 New AI Minds Podcast Episode with Founder and CEO of xAGI, Saurav Kumar!

• 📝 Free Transcription Tool for You!

• 👓 The World’s Greatest Mathematician discusses AI’s math abilities (Terence Tao)

• 🗞️ Why the New York Times is worried about Synthetic Data

• ❗ Bonus Content: AI Weirdness with Janelle Shane and More Mixture-of-Experts

• ♊ Glossary Page - Voice Cloning: Excitement and Fear

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

🎥  GPT-4 used to suck at math, but now…

This video summarizes a paper by Google DeepMind which introduces two new models—AlphaProof and AlphaGeometry 2— that can solve advanced reasoning problems (read: Olympiad-level problems) in mathematics. These results are unprecedented and nearly outperform human reasoning skills.

🧑‍🔬 Agentic AI: How to Build Them and What to Worry About

Building Living Software Systems with Generative & Agentic AI - This paper is an opinion paper that looks at the future of computing in the age of Generative and Agentic AI. That is, this paper argues that "Living software systems" powered by generative AI offer an alternative to the traditional, rigid and static software we’ve used for so many years.

Agent AI: Surveying the Horizons of Multimodal Interaction - This paper argues that by developing agentic AI systems in grounded environments, one can also mitigate the hallucinations of large foundation models and their tendency to generate environmentally incorrect outputs.

💻 Virtual Workshop: Building Voice AI Agents at Scale with Deepgram & Groq

Master building voice AI agents in this practical, hands-on workshop hosted by Deepgram and Groq. Limited time: get 20% off with code AUGAI20 until 08/31. Sign up

When: Friday, September 20th | 9AM - 12PM PT

Where: Zoom

🏇 The Best Text-to-Speech APIs and The Art of Finetuning LLMs

The Best Text-to-Speech APIs - With numerous options available, selecting the best Text-to-Speech API for your needs can feel overwhelming. Some APIs excel at producing high-quality voices, while others specialize in real-time conversational responses. This definitive guide will help you make the right TTS decisions for your use case.

Building an LLM Stack Part 3: The art and magic of Fine-tuning - This article is the final installment of a three-part series on how LLMs are built from scratch, and how you can make improvements to any OSS models out there. Whether you want to build a personalized chatbot or a copilot coder, this article will help you on your journey.

🐝 Social Media Buzz: Agentic AI that builds more Agentic AI

📲 Trending AI Apps for you!

AgentGPT allows users to configure and deploy autonomous AI agents with ease directly from a browser interface. By simply naming your AI agent and setting a goal for it to achieve, you initiate a process where the agent begins to ideate, execute tasks, and learn from outcomes.

ChatHub operates as a browser extension compatible with all Chromium-based browsers such as Chrome, Edge, and Brave. This all-in-one chatbot client—designed as a browser extension—facilitates the simultaneous utilization of multiple chatbots.

Aiorde leverages advanced artificial intelligence technology to provide users with an in-depth analysis of stocks and cryptocurrencies. By integrating AI with a user-friendly interface, the platform offers personalized stock filters and technical indicators that help investors refine their strategies and identify potential investment opportunities.

🎙️ AI Minds Podcast!

Saurav Kumar, Founder and CEO of xAGI, talks about his journey from creating video compression technology to starting xAGI, discussing its key AI products and sharing insights on AI development and entrepreneurship.

xAGI's AI health insurance advisor offers personalized, unbiased recommendations, showcasing AI's potential in healthcare.

📝 Free Transcription Forever! New Speech-to-Text AI Tool

Looking for a simple way to convert speech to text? Deepgram's free transcription tool is your ultimate solution. Whether it's conversations, audio files, or YouTube videos, our advanced AI transcription tool supports over 36 languages and dialects, making it the best free AI transcription tool available online. Discover how easy and efficient transcription can be with our tool.

🤖 Bonus Bits and Bytes!

If you've scrolled this far down, we've got some exciting bonus bits of content for you!

• World’s Greatest Mathematician discusses AI’s math abilities - Terence Tao, often described as the world’s greatest living mathematician, speaks at the 65th International Math Olympiad about Machine Learning.

• When AI’s Output is a Threat to AI Itself (NYT) - The New York Times is worried about synthetic data, as they say the following: “As A.I.-generated data becomes harder to detect, it’s increasingly likely to be ingested by future A.I., leading to worse results.”

• AI Weirdness - Janelle Shane’s genius and humorous blog strikes again as she tries to work with LLMs, image generation, and—most frustratingly—image manipulation.

• Mixture of Experts: How an Ensemble of AI Models Decide As One - (Featured last week) We’ve gotten numerous requests to feature this article once again on Mixture of Experts… perhaps you’ll find some new, updated information this time around as well 👀

• Voice Cloning: Excitement and Fear - This article delves deep into the heart of voice cloning—revealing not just what it is, but the groundbreaking science that powers it.