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:

• 🤖 Clone’s new, muscular robot looks more human than ever

• 📊 Why AI-reviewed papers might outperform peer-reviewed ones

• 👓 CasualBench: A Benchmark for Evaluating LLMs’ Casual Reasoning

• ☁️ Why Zuckerberg wouldn’t put upload his brain into the cloud?

• 🐦 Social Media Buzz: How and Why AI hurts students but helps adults

• 🤔 What exactly is an AI Agent, and are they actually useful?

• 🦙 Llama 3.1 and Mistral NeMo: A full review

• 🎙️ New episode of the AI Minds Podcast with CPO of Humach Bruce Sharpe!

• 📲 Three new, trending AI Apps for you

• 🎨 I made 60 FPS flipbooks using AI: How far we’ve come

• 🤣 Which AI language model is the funniest?

• 🚘 RAG vs. Finetuning: Enhancing LLMs with new knowledge

• 📚 What’s the Ego 4D Dataset and why is it revolutionary?

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

🎥  Brains over Brawn? A look at (non-intelligent) robotic muscles

This video demonstrates perhaps the next big leap forward in the world of robotics: AI Muscles. While most robots today seem mostly skeletal, Clone’s latest demo “Torso” includes an actuated elbow, cervical spine (neck), and anthropomorphic shoulders with the sternoclavicular, acromioclavicular, scapulothoracic and glenohumeral joints. The valve matrix fits compactly inside the ribcage. Bimanual manipulation training is in progress.

We haven’t paired it with an AI brain yet, but time may come sooner than you think…

🧑‍🔬  Peer-reviewed papers versus GPT-reviewed papers… Who Comes Out on Top?

Can large language models provide useful feedback on research papers? A large-scale empirical analysis - The authors of this paper created an automated pipeline using GPT-4 to provide comments on the full PDFs of scientific papers. Overall, more than half (57.4%) of the users found GPT-4 generated feedback helpful/very helpful and 82.4% found it more beneficial than feedback from at least some human reviewers

CausalBench: A Comprehensive Benchmark for Evaluating Causal Reasoning Capabilities of Large Language Models - Existing datasets for evaluating causal reasoning in LLMs are limited by narrow domain coverage. To address these shortcomings, this paper introduces a novel benchmark that spans textual, mathematical, and coding problem domains.

🐝 Social Media Buzz: AI Harms Students… but helps adults?

🏇 AI Expert Breaks Down AI Agents and Why LLMs May Be Stagnating

What exactly is an AI Agent? - In this article, AI expert Tife Sanusi covers what exactly an AI Agent is, how they work, and why so many companies (and individuals) are benefitting from them. If you’re on the fence about incorporating AI agents into your workflow or business, this article is for you!

Improvement or Stagnant? Llama 3.1 and Mistral NeMo - With the rapid development of LLMs, the emergence of new models boasting improved performance has become the norm. Two well-known open-source series, Llama and Mistral, have recently introduced new stars to their lineups: Llama 3.1 and Mistral NeMo. But are these models truly improvements on their previous iterations? Or are returns visibly diminishing?

🎤 The AI Minds Podcast!

In this episode of AI Minds, host Demetrios speaks with Bruce Sharpe, Chief Product Officer at Humach, about his journey through telecommunications, AI, and customer experience innovations. From developing omni-channel platforms at Ticketmaster to pioneering AI-powered voice interactions, Bruce shares his insights into the critical intersection of human and machine interactions in regulated industries like healthcare and insurance.

📲 Trending AI Apps for you!

Marblism is an innovative platform designed to streamline the process of launching full-stack NextJS applications quickly and efficiently. This tool is a game-changer for developers, offering a robust solution to create and deploy applications in a fraction of the time traditionally required. With Marblism, users can focus on adding their unique features rather than getting bogged down with initial setup and configuration tasks. 

Papers, a leading reference management solution, offers an advanced system designed to streamline research organization and discovery. Trusted by millions globally, this tool is not only popular among students but is also widely used in academic and corporate environments. With Papers, users can seamlessly manage references, enhance collaboration, and maximize research efficiency.

SoundAI leverages cutting-edge artificial intelligence technology to transform users' musical concepts into auditory reality. The platform simplifies the music creation process by providing a suite of AI-powered tools that automate and enhance sound generation. By utilizing AI algorithms, SoundAI can interpret user inputs such as textual descriptions or uploaded audio files and generate corresponding musical outputs in real-time.

🤖 Bonus Bits and Bytes!

If you’ve scrolled through this far, please enjoy some bonus content curated especially for you!

• 🎥 I made 60fps Flipbooks using AI - This video is two years old… before Sora and before the latest iterations of Stable Diffusion. How good (or bad) is the quality of this art? Find out here!

• 😂 Which AI language model is the funniest? - This article by Ben More of The Carletonian puts various LLMs to the comedic test. The winner may shock you!

• 🧠RAG vs. Finetuning: Enhancing LLMs with new knowledge - ML Engineer Brad Nikkel discusses the differences and advantages of RAG and Finetuning in this in-depth article.

• 📊 Ego 4D Dataset - With an ever-growing digital universe, the challenge of capturing, storing, and making sense of web data has never been more critical. Enter the Ego4D Dataset: a monumental collection that stands at the forefront of this exploratory frontier.