Tesla builds largest AI supercomputer

Tesla has assembled a monster AI computer from 5,760 of Nvidia’s high end Graphical Processing Units (GPUs), capable of 1.8 quintillion (1.8 million million million) complex operations per second - the largest in the industry. All of this firepower has been assembled to train AI models for self-driving cars. Elon has announced that Tesla’s next-generation self-driving software will be trained on the video data from the 8 cameras on each of the 4 million existing Tesla vehicles in service - with NO human generated code. AI, all the way down.

Tesla’s monster AI supercomputer

Amazon turns your palm into a payment device

No cash, no card, no phone? No problem. Amazon has developed a system that identifies you by scanning your palm print, allowing for secure payments linked to your bank account or credit cards. The system incorporates an AI model trained on several million artificially-generated images of palms, to allow it to recognize the unique lines, grooves, and ridges on each palm, as well as the network of pulsating veins beneath the skin. Amazon claims that the system has demonstrated 99.9999% accuracy, 100 times better than current iris-scanning technology. Coming soon, to a Whole Foods near you.

Amazon palm print payment system

The secret history of OpenAI’s rise to AI dominance

Ever since ChatGPT became the fastest-growing consumer app in history last November, rocketing quickly to over 100 million users, OpenAI has been the 800-pound gorilla of generative AI. This week, Wired magazine unveils the little-known backstory of how a ragtag band of passionate visionaries stumbled towards greatness and created, almost in spite of themselves, what may become one of the most consequential industries in history.

OpenAI wants to change everything. Yes, EVERYTHING.

Digital avatar Noonoouri “signs” record deal with Warner Music

A digital avatar with 400,000 Instagram followers has landed a contract with Warner Music to release “her” first single. Noonoouri, creation of graphic designer Joerg Zuber of Munich, Germany, has been a social media influencer since 2018. She has been a spokesmodel for a number of leading luxury brands, including Dior and Kim Kardashian’s KKW Beauty. Now, with the help of AI, she will have a digital singing voice.

Digital avatar lands record deal

Tiny island nation Anguilla gets rich from selling “.ai” domains

AI is hot. Which makes domain names ending in “.ai” highly sought after. This has been excellent news for the tiny island nation of Anguilla, a mere 35 square miles of beach and sand, with fewer than 20,000 inhabitants. Its ownership of the top level domain “.ai” is projected to add up to $30 million dollars to its coffers this year, representing approximately 10% of GDP.

Anguilla makes bank from “.ai”

Indian company claims AI powered anti-drone defense system can protect an entire city

The war in Ukraine has shown the devastating effectiveness of drone attacks, which are increasingly employed by both sides. Drones are small, fast, agile, self-guided by onboard AI systems, and therefore hard to predict, detect, and track with conventional radar or laser defenses.

Now Hyderabad-based Grene Robotics announces an AI-powered autonomous anti-drone defense system than can protect an entire city.

Details are scanty at this point, but the need is great, and competitors are sure to rush into this space; even if Grene Robotics falters. Set an AI to catch an AI - this is part of the future of warfare.

AI anti-drone defense system

New text-to-video startup nips at the heels of Runway

Pika Labs, a text-to-video startup founded by two Stanford PhD students 5 months ago, has VC’s so eager to invest that the company is on the verge of completing their third funding round in the last 5 months. Its Discord server has grown from 0 users to over 160,000 in 2 months. It even has category leader RunwayAI running scared, copycatting Pika’s new features soon after the company adds them. More competition is in the wings: leading text-to-image startup Midjourney is toying with entering the text-to-video space soon.

See what the excitement is about with these demos:

Clip 1 Clip 2

Join the beta here

(Hat tip to Stephanie Palazzolo of the AI Agenda)

UK startup uses AI to detect lung cancer early

Lung cancer is the leading cause of cancer-related deaths. Founded by Oxford University researchers, Optellum uses AI image analysis to help classify nodules found in lung scans into high or low risk for cancer. Studies have shown that the system improves diagnosis significantly for both generalist Pulmonologists and Radiologists, as well as for experts in the diagnosis of the disease. This allows for earlier detection, when intervention has a greater chance of cure. The system has been cleared by the US FDA, and is in use in hospitals in both the US and the UK.

AI-powered hearing aids adapt autonomously to noisy environments - and translate foreign languages on the fly

Minneapolis-based hearing aid company Starkey is embedding AI into its new Genesis AI devices to unlock a host of extra features, including automatically adjusting settings to optimize hearing for speech in noisy environments, and even translating a foreign language in real time. Now your grandkid can wear these hearing aids and finally get an A in that French class he’s failing.

High tech hearing aids

ChatGPT can complete surgical notes faster and better than human surgeons

Documentation is the bane of a busy doctor’s life. Plastic surgeons at Queen Victoria Hospital in the UK studied the effectiveness of using GPT-4 to complete surgical notes after a procedure. They found that GPT-4 completed notes far faster (averaging 5.1 seconds compared to the surgeons’ 7.1 minutes) and better (as rated by both surgeons and patients.)

GPT-4 completes surgical notes faster and better than surgeons