👽 NASA wants to use AI to study UFOs

🎨 AI Image of The Week

🧠 AI and Machine Learning

Will AGI Be Built in China? (4 minute read)
China's Baichuan Intelligent Technology has unveiled two impressive open-source large language models, named Baichuan 2-7B and Baichuan 2-13B. What got everybody talking was their stellar performance on AGIEval, a benchmark by Microsoft Research. Baichuan 2-13B managed to one-up ChatGPT with a score of 48.17, while ChatGPT trailed at 46.13. The secret sauce? Baichuan 2-13B was trained on a treasure trove of Chinese-English bilingual data, pulling insights from millions of webpages covering a wide range of positive domains, from policies and laws to traditional values and more. Delving into AGIEval's research paper, we discover its broad scope, encompassing not just SAT and LSAT but also Chinese exams like Gaokao and bilingual tasks, truly putting these AI models to the test in multiple languages and domains.

AI’s Next Leap: From Trillions Of Tokens To Every Word Ever Written (3 minute read)
At MIT's AI for Impact course, attendees got a front-row seat to the astonishing trajectory of AI's evolution. Picture this: a whopping 10,000-fold surge in computer resources dedicated to AI in just one decade! It's not just growth; it's an all-out revolution, following the pace set by Moore’s Law, where processing power doubles every two years. But that's not all—the fuel for these AI engines is trillions of tokens, a staggering amount. To put it in perspective, think about the entire Library of Congress, a vast reservoir of human knowledge, totaling about 20TB of text. The ambitious minds driving AI are setting their sights on encompassing effectively every word ever written in human history for the next training model.

💼 Business

Jensen Huang's Unusual Nvidia Management Style Doesn't Include Long-Term Plans or Reports (4 minute read)
In a spectacular feat, Nvidia has claimed the title of the world's most valuable semiconductor company, boasting a staggering value exceeding $1 trillion, outpacing giants like Intel. But whilst this isn’t necessarily new news, people are talking about the unconventional management style of Nvidia's CEO, Jensen Huang. Despite steering a company that earns more than its industry counterparts, Huang's approach is refreshingly distinct, void of long-term planning and featuring a remarkable 40 direct reports. Recently, in an interview with Joel Hellermark, he offered a glimpse into his unique management philosophy, unveiling a flat organizational structure that shuns traditional hierarchy and embraces inclusive group discussions over solo tête-à-têtes, ensuring a level playing field of communication right from the top down.

Arm after the IPO (3 minute read)
"The future of AI is intrinsically tied to the growth of Arm," remarked Arm EVP and Chief Commercial Officer Will Abbey, as Arm's stock made its Nasdaq debut. While Arm might not be the first thing that pops into your head when you think of AI, Abbey is quick to connect the dots. He emphasized that AI already operates on Arm today, and as AI continues its ubiquitous spread, the demand for enhanced computing capabilities, heightened power efficiency, and AI-centric software ecosystems will drive Arm forward. Abbey even gave a nod to Nvidia's Grace Hopper superchip, where Arm's Neoverse cores and Nvidia's H100 Tensor Core GPU collaborated harmoniously.

❓ Misc

NASA wants to use AI to study unidentified aerial phenomenon (2 minute read)
A recently commissioned independent study, backed by NASA, suggests combining NASA's expertise and public trust with artificial intelligence to investigate unidentified aerial phenomena (UAP) on Earth. In response, NASA Director Bill Nelson has appointed a NASA Director of UAP Research to lead and oversee these investigation efforts. Nicola Fox, associate administrator of the Science Mission Directorate at NASA, emphasizes the importance of this role in providing guidance and operational coordination to identify and understand these puzzling aerial phenomena. While NASA clarifies there's no evidence of extraterrestrial origin, they admit to not having a definitive explanation for some documented UAP sightings. The aim is to shift the focus from sensationalism to science, using open-source UAP data to keep the public informed.

Australian, UK and US tech companies already reaping AUKUS benefits (5 minute read)
Artificial intelligence and autonomy companies hailing from Australia, the United Kingdom, and the United States are in a race to develop and promote tools that can gather extensive data and assist operators in navigating information-rich environments. Their objective is to secure contracts both domestically and with allied nations, particularly as details about the second phase of the AUKUS trilateral arrangement, focused on advanced technology, emerge this fall. According to Cross, Vantage, an AI-powered decision aid, is a crucial tool for future allies. Drawing from a wide array of in-house and open-source documents, Vantage continuously builds its information base. For instance, it can respond to queries like, "Given the increasing Russian influence in Mali, what threat does this pose to the region's stability?" showcased during a September 13 demonstration at DSEI by Ollie Carmichael, the product manager. Adarga has already extended Vantage to its existing military, government, and commercial customers, and recently made it accessible to a broader customer base.

Thanks for reading, if you enjoyed, tell your friends!