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:

• 🎤 Keynote: How Karpathy built AI Recommendation in a Weekend

• 🧪 Chemical General Intelligence: An LLM Demystifies Chemistry

• ⌨️ The World’s Smartest Keyboard (or Gboard)

• 🌊 Liquid Neural Networks: Everything you need to know

• 💰 AI in Finance: Literal Money Making Machines

• 🐦 Social Media Buzz: AI-generated proteins and RL Robots

• 📲 New Trending AI Apps for You!

• 🎙️ AI Minds Podcast with CEO Ross Young

• 🔥 The Official PyTorch Documentary

• 🔋 How AI caused a shock to Material Science

• 🏆 Updated! The Absolute Best Speech-to-Text APIs in 2024

• 🎲 Monte Carlo Learning: Predictions about Predictions

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

🎥  How Andrej Karpathy Built an AI Recommendation System in a Weekend

At a hackathon in UC Berkeley, founding member of OpenAI Andrej Karpathy gave an incredible keynote. He goes over his story, from building game-playing programs to an AI recommendation engine. To find out how he built such complex ML models in a mere weekend, watch this video!

🧑‍🔬  Chemical General Intelligence and The World’s Smartest Keyboard

ChemDFM: Dialogue Foundation Model for Chemistry - You’ve heard of Artificial General Intelligence (AGI), but what about Chemical General Intelligence (CGI)? The existence of specialized language/knowledge in the field of chemistry hinders the performance of general-domain LLMs in chemistry. To this end, this paper presents ChemDFM, the first LLM towards CGI—demystifying how chemicals work for everyone.

Proofread: Fixes All Errors with One Tap - This paper demonstrates Proofread, a novel Gboard feature powered by a server-side LLM, enabling seamless sentence- and paragraph-level corrections. To obtain quality models, the authors employ a two-stage tuning approach: Supervised Fine Tuning for foundational quality, followed by Reinforcement Learning for targeted refinement.

🏇 Liquid Neural Networks and AI Advancements in Finance

Liquid Neural Networks: Fluid, Flexible Neurons - Liquid neural networks improve upon traditional neural nets in unprecedented ways. With dynamic neurons and a suite of optimizations, liquid neural networks are pivotal to understand to keep up with the AI world. Here’s all you need to know.

AI in Finance: Money Making Machines - In 2016, long before the mainstream popularization of AI, JP Morgan Chase was looking for ways to boost their capabilities as a firm. Sure enough, they turned to Machine Learning. Here’s what they learned.

🐝 Social Media Buzz

📲 Trending AI Apps for You!

DeepAI is an accessible suite of AI tools designed for artists, writers, and designers. Its features include an AI image generator and editor, AI chat bots, and an AI search engine, making AI creativity easily attainable.

Transistor.fm serves as a user-friendly guide in navigating the complexities of podcasting. Users can record and upload their audio, and Transistor takes care of the submission and distribution across platforms like Apple Podcasts and Spotify. 

Vizologi is an AI-powered software that helps businesses generate ideas, gain market insights, analyze competitors, and automate business plan creation. From brainstorming to getting data-driven insights on markets, trends, and competitors, Vizologi is a fascinating and efficient tool for you and your team!

🎙️ AI Minds Podcast!

In this episode of AIMinds Podcast, Ross Young, CEO at Clinical Notes AI, shares his journey from the cybersecurity sector to revolutionizing clinical notes AI for behavioral health.

Ross dispels common myths about AI replacing jobs by showcasing how AI can be a powerful tool to enhance human capabilities—creating tools that optimize efficiency, improve healthcare outcomes, and ultimately, make a positive impact on people's lives.

🤖 Bonus Bits and Bytes!

• The Official PyTorch Documentary - This film unveils the authentic narrative of PyTorch’s inception, attributing its existence to a dedicated group of unsung heroes driving technological innovation.

• AI-enabled materials discovery for advanced ceramic electrochemical cells - AI might help us develop new methods of clean and efficient energy conversion. Material Science is indeed getting a boost from machine learning models. Here’s how.

• The Best Speech-to-Text APIs in 2024  - The title says it all. If you want to build a voice AI agent or generate subtitles for your podcast, check out these tools!

• Monte Carlo Learning: Everything You Need to Know - Machine learning comes in many flavors, and if you’re interested in predictions about predictions (Prediction-ception?), check out this in-depth guide into Monte Carlo Learning!