Nvidia: World’s most valuable company

At first serving gamers, then crypto-bros, and now AI, chipmaker Nvidia has stair-stepped its way into becoming the public company with the highest market capitalization in the world. The company was created in 1993 to bring better graphics to computer games - and initially failed miserably, because its original tech just didn’t work. On the edge of bankruptcy, the young company fought back and eventually dominated the graphics cards needed for high end games. Then their graphic chips turned out to be really good at “mining” (creating) bitcoins and other crypto currencies, and the company grew rapidly in that volatile market. AI came to the forefront in 2022 and Nvidia now produces 95% of all Graphic Processing Units (GPUs) used in the AI industry. Recently, Nvidia passed $3 trillion (with a “T”) in market capitalization, joining Microsoft and Apple, the only other companies in this lofty tier. And now, Nvidia sits at the top of the heap, headed by an unlikely working class hero, CEO Jensen Huang, an immigrant who once cleaned toilets for a living. Even if Nvidia’s supremacy is fleeting, this is an “only in America” story that deserves wider notice.

Nvidia becomes the world’s most valuable public company

Search monsters battle for supremacy

Google is redesigning its search engine, and it’s AI all the way down. Despite having invented modern AI in 2017, Google has been slow to commercialize the technology. Now though, it seems to be committed to rebuilding its flagship search technology to incorporate AI at all levels. The challenge for them will be to keep or match the torrents of ad money that the old Google search delivered, with its scads of links in the results. AI-produced summaries may not engender the profitable ad click-throughs that Google has been accustomed to, which explains Google’s ambivalence up to now.

Upstart Perplexity, the “AI-native” search product often touted as the successor to tired old Google, is fighting back by putting little widgets for weather, time, sports scores, etc. on the search page. Perplexity has built a loyal following by being much better at research and complex queries than link-obsessed Google. Still, sometimes a girl just wants to know the temperature outside, simple and quick, not a damn Wikipedia page on the history of weather forecasting. To dethrone Google, Perplexity needs to be able to handle both.

Google redesigns its search engine, and it’s AI all the way down

AI startup Perplexity adds widgets so users don’t need Google

US Navy uses AI-enhanced underwater drones to find mines

AI models loaded onto underwater drones have mostly replaced time- and manpower-intensive manual reviews of underwater drone footage to distinguish fish traps from mines. The resulting assessments are delivered much faster. The AI models can also be continuously updated over radio link when on the surface, allowing for adaptation to different seabed characteristics or different camouflage by adversaries. The trials have been so successful that a substantial expansion of the program is in the works.

US Navy uses AI-enhanced underwater drones to detect mines

UK AI candidate vows to “humanize politics”

“AI Steve”, an AI persona represented by Sussex businessman Steve Endicott, will appear on the ballot in the upcoming UK elections for Parliament. AI Steve is running to represent voters in Brighton Pavilion, a community on England’s south coast. Endicott says the idea is to create a politician that is always available to answer questions from constituents, and that can take their views into account. He has already created a voice-enabled AI chatbot that has been answering hundreds of questions per day from constituents. Whether this is more than a publicity stunt for Endicott’s voice-assistant company, English voters will have to decide.

UK AI candidate vows to “humanize politics”

AI helps preserve Amazon rainforest

The Amazon is the world’s largest rainforest, sometimes described as the “lungs of the planet” because of the carbon-sequestering and oxygen-producing activities of its trees and plants. Unfortunately, it is currently being deforested at an alarming rate due to encroachment from illegal farming and mining operations. Now AI is helping researchers and governments monitor deforestation activities through networks of cameras and sensors. This helps governments avoid playing whack-a-mole with lawbreakers who slip quickly from one area to another to avoid detection or apprehension.

AI helps preserve Amazon rainforest

Optical chips may solve the AI power problem

The projected power requirements for AI in the future are staggering. The International Energy Agency predicts that AI will consume 10 times as much electricity in 2026 as it did in 2023, and that data centers in that year will use as much power as all of Japan. Since necessity is the mother of invention, much attention is being focused on how to avoid an AI energy crunch. One promising technology that could significantly decrease the AI energy drain is to switch from computer chips that use electrons for calculations, to chips that compute with light. Researchers at MIT have demonstrated optical computers that can run small AI models using 1000 times less energy than conventional electronic computer chips. The technology is already being commercialized, and venture capital is funding several startups. Kicking electronic chips out of the data centers is a distant dream for now, but progress in the field is rapid.

Optical chips may solve AI’s power problem

UK company uses AI for rapid development of “green” motor

One of the problems with transitioning to a green energy future is that batteries and magnets use a lot of rare earth elements that are, well, rare (and therefore expensive) and highly toxic. Development of high-performing alternatives that use more conventional materials that are cheaper and safer has been up to now a lengthy trial and error process. A UK tech company announced last week that they had used AI to design, synthesize, and test a promising “clean earth” alternative magnet for electric motors in only 3 months. AI seems particularly well suited for materials science, with some of the most promising AI startups in the world focusing on new drug discovery, a field where tiny amounts of a new molecule can be worth billions of dollars.

UK company uses AI for super-fast development of “green” motor

AI blood test detects Parkinsons 7 years before any symptoms

As the world population ages, the prevalence of Parkinson’s Disease is growing rapidly, currently affecting over 10 million people worldwide, and found in approximately 2% of all individuals over the age of 80. Early detection has been difficult, and most people are not diagnosed until they have notable symptoms, at which point the disease is already well advanced. A team of researchers in the UK, Germany, and Italy have used an AI model to devise a blood test that analyzes the expression of 8 proteins in the blood, to identify at-risk individuals who may be candidates for clinical trials of interventions aimed at slowing or halting progression of the disease.

AI blood test detects Parkinsons 7 years prior to symptoms

OpenAI announces cancer collaboration with Color Health

OpenAI has announced a collaboration with cancer care company Color Health, to develop an AI Copilot to help doctors accelerate patient access to optimized, personalized cancer care. Color Health points to the delays faced by cancer patients in the diagnosis and treatment planning phases of their care. They estimate that a month’s delay in diagnosis can notably decrease the chance for an optimal outcome of treatment. Color wants to use AI to accelerate the process of arriving at the correct diagnosis, and then speed up development of an optimized, personalized treatment plan. The model is “human in the loop”, where the AI assists the physician with the details of the care process, rather than replacing the human.

WSJ: OpenAI expands into healthcare with Color Health cancer copilot

OpenAI Announcement: Helping doctors transform cancer care

UMass Memorial speeds ER triage with “AI Kate”

Nurses at UMass Memorial Hospital in Worcester, Massachusetts have been using an AI tool from software vendor Mednition for the past year, to help them accurately classify the acuity of ER patients according to the standard Emergency Severity Index. Nicknamed “Kate”, the AI system assesses the patient information entered by the nurses and suggests an index score. An accurate score is key to allocating scarce and time-sensitive ER resources to the patients most in need of them.

UMass Memorial speeds ER triage with AI “Kate”

AI detects cancer DNA in a simple blood test

A worldwide team of researchers, including some from the Massachusetts General Hospital in Boston, have used AI to develop an ultra-sensitive blood test for circulating tumor DNA. This test, known as MRD-EDGE is being used initially to assess the presence and quantity of residual disease in cancer patients that have undergone one or more phases of treatment. This allows better planning for future interventions.

AI detects cancer DNA in simple blood test