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:

• 🤖 Video: Boston Dynamics’ robot “Atlas” is now fully autonomous!

• 🦃 LLMs’ problem with Thanksgiving

• 😂 Humor in AI: Massive Scale Crowd-Sourced Preferences and Benchmarks

• 🐦 Social Media Buzz: Sam Altman amends a Tweet from two years ago

• 🧠 Karpathy discusses cool GPT experiment using months of prompts and memory

• 📈 Anthropic CEO says unless something goes wrong, we’ll have AGI in 2026/2027

• 💻 New Webinar! “Beyond the Cloud: Securely Harnessing the Power of Voice AI”

• 🎮 Tutorial and Demo: How to build a Voice AI Video Game in Python

• 🎤 AI Minds Podcast with Michael Ocean, CEO and Cofounder of SellMeThisPen

• 🧨 The Threat of AI Weapons, as predicted by Veritasium

• 🚖 You might’ve missed this announcement from Tesla!

• ⌨️ A full rundown of AI Hardware: Glossary Entry

• 🚀 How to build Star Trek’s universal translator with current LLMs and ML Models

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

🎥  Boston Dynamics’ Atlas is now fully autonomous!

Description: 

“In this video, Boston Dynamics’ robot Atlas is autonomously moving engine covers between supplier containers and a mobile sequencing dolly. The robot receives as input a list of bin locations to move parts between.

Atlas uses a machine learning (ML) vision model to detect and localize the environment fixtures and individual bins [0:36]. The robot uses a specialized grasping policy and continuously estimates the state of manipulated objects to achieve the task.

There are no prescribed or teleoperated movements; all motions are generated autonomously online. The robot is able to detect and react to changes in the environment (e.g., moving fixtures) and action failures (e.g., failure to insert the cover, tripping, environment collisions [1:24]) using a combination of vision, force, and proprioceptive sensors.”

🧑‍🔬  LLM’s Problem with Thanksgiving and the Latest Benchmark in AI Humor

Not All Countries Celebrate Thanksgiving: On the Cultural Dominance in Large Language Models - With November here, we thought it would a fun idea to highlight this academic paper. The authors identify a cultural dominance issue within large language models (LLMs) due to the predominant use of English data in model training. To systematically evaluate the cultural dominance issue, the authors build a benchmark of concrete (e.g., holidays and songs) and abstract (e.g., values and opinions) cultural objects. Here’s how GPT-4 performs!

Humor in AI: Massive Scale Crowd-Sourced Preferences and Benchmarks for Cartoon Captioning - This paper presents a novel multimodal preference dataset for creative tasks, consisting of over 250 million human ratings on more than 2.2 million captions, collected through crowdsourcing rating data for The New Yorker's weekly cartoon caption contest over the past eight years. With it, you can finetune and test your models for humor. Check it out!

🐝 Social Media Buzz: Altman’s amendment and Karpathy on AI’s months-long memory

🏇 New Webinar on Voice AI Solutions and 

Voice AI adoption is surging, but balancing security and compliance is critical, with 83% of enterprises moving workloads to private clouds for greater control. Join our webinar on November 19th to explore the advantages of on-premises Voice AI solutions and how to safeguard sensitive voice data and ensure compliance, without compromising real-time performance. Reserve your spot today!

When: Tuesday, November 19th at 10AM PT | 12 PM CT | 1 PM ET

Where: Online

Register here!

Tutorial and Demo: How to Build a Voice AI Video Game in Python - In this article, AI researcher Andy Wang will teach you to transform an ordinary platformer game into one that can be controlled by your voice using Deepgram’s API. Note: the focus here is more on how Deepgram can be integrated into the mechanics of a game and less on how to program a platformer in Python. Check it out!

🎤 The AI Minds Podcast!

Michael Ocean, a 10-year sales veteran and former leader at Hopin, shares his transition from musician to entrepreneur. As the founder and CEO of SellMeThisPen, he discusses how his early startup experiences and time at Hopin influenced his approach to business. Noticing issues with traditional sales training, he started SellMeThisPen to provide real-time, personalized training aimed at improving sales team performance.

🤖 Bonus Bits and Bytes!

If you’ve scrolled down this far, here are some bonus features for you!

• 🎥 The Threat of AI Weapons - Veritasium, an undeniable pioneer in the world of YouTube education, created a video six years ago on the consequences of AI Weapons. Do you think his arguments still hold true?

• 🚕 Slept on: The Cybercab - With major events like the US election and other technological advancements all taking place simultaneously, you might’ve missed the announcement of Tesla’s Cybercab! Check it out here

• 💾 A full rundown of AI Hardware: In this glossary entry, you’ll learn everything you need to know about GPUs, TPUs, and NPUs—from the chips in your phone to the innovations of NVIDIA. 

• 🚀 Sci-Fi Tech Come to Life: Building the Universal Translator from Star Trek with AI and LLMs - In this article and video, AI engineer Jose Francisco will teach you to build a universal translator with AI, just like the ones in Rick and Morty, Star Trek, and the Marvel Cinematic Universe.