Delaware court nixes Elon’s pay package

Expect MAJOR whining. Elon Musk, the world’s richest man-baby, has been told “No!” by a Delaware court, which disallowed his 2018 mega-pay package because it was a sweetheart deal negotiated with a board composed of all his BFFs. Delaware is a notoriously corporation-friendly state, but this $56 billion pay package, the largest in corporate history, was such a blatant giveaway of corporate value to a single shareholder that even the Delaware judge gagged. Elon recently complained on Twitter/X that he is still underpaid (?!?!?), and he needs at least double the amount of shares he now has (currently worth $205 billion.) See below for his “pouty-face.”

Delaware court nixes Elon’s sweetheart pay deal

Robots are fighting robots in Russia’s war in Ukraine

You’ve heard about the aerial drone attacks in the war in Ukraine. Now ground vehicles drones are hauling ammunition, clearing mines, evacuating the wounded, and carrying radio-jamming equipment to knock drones from the sky. Most of these robots are controlled remotely by humans, but the next generation will be more autonomous, and more capable of robot-on-robot combat.

Robots fight robots in Russia’s war in Ukraine

Blackstone is building a $25 billion empire of data centers

Data is the new oil. And Blackstone Group, the trillion-dollar private equity behemoth, is planning to handle gushers of it.

Blackstone is reportedly planning to pour up to $25 billion into building out giant data centers around the globe. The largest will be millions of square feet, and consume hundreds of million Megawatts of electricity, enough to power hundreds of thousands of homes. All this to meet the ravenous demand of businesses for cloud computing, and compute-intensive AI. Rural spaces that have access to reliable electricity are being transformed into acre on acre of compute farms. Clashes with environmentalists are already flaring.

Blackstone turning farmland to server farms

+

Samsung plans all-AI, no-human chip factories by 2030

Humans are messy and unreliable. Making computer chips requires nanometer-scale precision. Korean electronics giant Samsung has decided that the time has come to make chipmaking factories completely run by AI, and expects to make this a reality by 2030. It’s like the old joke: “Factories of the future will have only 2 employees, a human and a dog. The job of the human is to feed the dog. The job of the dog is to make sure that the human doesn’t touch the machines.”

Samsung plans all-AI, no-human chip factories by 2030

AI robot startup Figure in talks with Microsoft and OpenAI  

AI robot startup Figure is looking to raise up to $500 million, and is reportedly in talks with Microsoft and OpenAI to lead the funding. Figure has been getting increasing attention lately as its AI-powered humanoid robot has shown the capability to learn tasks from brief demonstrations by humans, even finicky household chores like making coffee.

AI robot startup Figure in talks to raise up to $500 million

Researchers develop AI “fingertip” that reads Braille

Researchers from the University of Cambridge have developed a robotic sensor that uses AI-augmented touch and vision to read Braille twice as fast as humans. This is merely a demonstration of the technique, opening the door to more complex uses of robotic touch, such as surgery.

Researchers develop AI “fingertip” that reads Braille

Arc Search app combines search and AI into something new

The Browser Company has just released a new search app on IOS that combines search and AI in a new way, delivering answers rather than links. Although not a fully-polished product yet, it does rethink how we will use the web for gathering information, and points the way to an AI-centric browsing experience. If these guys don’t cross the finish line, they are planting seeds for the ones who will. Wading through links to find your information, a la Google search, is not yet dead, but it’s dying.

Arc Search app debuts on IOS

AI generates even better ideas with better prompting

Many studies have proven that chatbots can come up with creative ideas that are equal to or better than those of most humans. However, in most studies, the humans come up with more diverse ideas. The chatbot responses tend to cluster around a few major themes. A new study indicates that with better prompting, chatbots can improve their creativity and rival the best humans. Help your chatbot unleash its inner Picasso or Leonardo - improve your prompting game!

AI generates more creative ideas with better prompting

MIT and Google develop AI to predict health from wearable data

Millions of Americans generate health data every day from their Fitbit or Apple Watch - and almost all of it is wasted. Now researchers from MIT and Google have developed a fine-tuned LLM based on the Alpaca version of Meta’s LLaMa, called Health Alpaca (oddly enough), which bests previous AIs’ performance in predicting health outcomes from wearable data. Mining the wealth of patient-specific physiologic data generated by wearables should be a boon for both population research on health outcomes, as well as targeted predictions and interventions for individuals.

MIT, Google develop AI to predict health outcomes from wearable data

Musk announces first human brain implant of Neuralink chip

Neuralink, an Elon Musk startup that is developing chips for direct implantation into human brains for a variety of medical purposes, has apparently placed its first chip in a human brain. Characteristically, Musk announced the event in a cryptic Tweet. This achievement appears to be part of a clinical trial to test methods for helping quadriplegics move again. Click the link for a thoughtful analysis by Scientific American.

Neuralink announces first human brain implant of its chip

Study: AI can help analyze text-based “psychotherapy at scale”

A recent article in JAMA reports on a study in which researchers used AI to analyze more than 20 million text-based counseling sessions, and were able to successfully predict both patient satisfaction and clinical outcomes. This points the way toward methods to analyze and maintain quality control in “psychotherapy at scale,” as well as gain new insights into what methods do and do not work in such online settings, which are a growing part of the psychotherapy landscape.

Overview of the study and its implications

The JAMA paper

GPT-4 beats Stanford Medical Students on patient case analysis

In yet another recent JAMA paper, OpenAI’s GPT-4 performed better on Stanford Medical School’s first- and second-year clinical reasoning exams than the Stanford medical students.

A quick overview on Twitter/X

The JAMA paper on clinical reasoning of chatbots vs students