Microsoft resurrects Three Mile Island nuke plant to power its data centers

The AI Age of tomorrow will require staggering amounts of electricity. At a recent White House conference on AI, Sam Altman, CEO of OpenAI and poster boy for the AI revolution, called for the US government to support the building of a multi-state network of 5 Gigawatt(!!) data centers (each using enough electricity to power a city of 3 million households) in order to keep the US economically strong and technologically ahead of the Chinese. As one small but telling part of this overall story, cloud computing providers like Microsoft and Amazon are turning back to nuclear power as a way to produce the needed electricity without burning fossil fuels. Microsoft has just announced plans to fund the revival of the shuttered Three Mile Island nuclear power plant in Pennsylvania, notorious as the site of the worst nuclear power disaster in US history, in return for a 20-year contract to use the electricity it will generate. For a company as averse to bad publicity as Microsoft to fund opening one of the iconic symbols of the dangers of nuclear power, the imperative to generate as much non-fossil-fuel energy as possible must be very, very strong indeed.

Microsoft resurrects notorious nuclear plant to power its data centers

OpenAI CEO sees superintelligence in “a few thousand days.”

Sam Altman posted a think piece to his blog, titled “The Intelligence Age.” In it, he paints a picture of a future world where artificial intelligence has created unimaginable abundance for everyone. The quote that has grabbed the most attention is his statement that achieving artificial super-intelligence (intelligence greater than any human) is possible in “a few thousand days.” That is, in as few as 3 to 10 years. Sam Altman is at the center of the AI revolution, talking to most of the smartest people in the field, and with one of the world’s great AI research labs literally within his own company. So if Sam thinks artificial super-intelligence is near, we need to pay attention.

Sam Altman thinks artificial super-intelligence is possible in “a few thousand days”

Microsoft partners with Palmer Luckey to turn US soldiers into Starship Troopers

Palmer Luckey is the college sophomore who invented a Virtual Reality headset that he turned into a business he sold to Facebook 2 years later for $2 billion. Ten years later, the billionaire boy genius has a new defense-tech startup called Anduril. Now Microsoft is teaming up with Luckey and Anduril to upgrade its IVAS headset for combat soldiers, which includes thermal and night vison into a heads-up display. The new IVAS will be a mixed-reality headset, allowing soldiers to see the reality before them overlaid with computer-generated information, such as threats beyond their line of sight.

Microsoft partners with Palmer Luckey to turn US soldiers into Starship Troopers

Nvidia and Alibaba partner to put AI into Chinese autos

AI chip-making powerhouse Nvidia is partnering with Chinese e-commerce giant Alibaba to bring AI to Chinese automakers, enabling autonomous driving as well as voice assistants capable of taking complex instructions from the driver, as part of an “intelligent cockpit.” Nvidia technology is already incorporated into the cars of many of China’s world-leading electric vehicle makers, and this partnership allows the EV automakers to add Alibaba’s highly capable Qwen AI models into the mix. Biden’s White House may have forbidden Nvidia to sell its top AI chips to China, but Nvidia is finding ways to do a robust business in that country, nonetheless.

Nvidia and Alibaba partner to put AI into Chinese autos

Fan fiction has turned to AI video mashups

Fan fiction, or fanfic, is made up of the creative works of fans of certain books, movies, TV series, or videogames, in order to extend the universe of the beloved original. Twilight, Star Wars, Hunger Games, Lord of the Rings, and many more works have thriving fan fiction communities online. Now AI has taken fan fiction in a new direction, with fans using new AI-powered text-to-video tools to remix old favorites into new, and sometimes very odd, video mashups. Click the link below to see a clip of “Harry Spotter, the Boy who Lifted,” a bodybuilder’s version of the Potterverse.

Fan fiction has turned to AI video mashups

Study shows AI adoption faster than PCs and the internet

A recent study from Harvard, Vanderbilt, and the Federal Reserve Bank of St Louis has found that generative AI, such as ChatGPT and its competitors, has infiltrated US society to a remarkable degree in less than 2 years. The study reveals that AI is being used to some extent by approximately 40% of Americans. Adoption of the internet took over 4 years to hit the 40% mark, and adoption of PCs required over 12 years to reach the same penetration. And uptake of AI is found in all sectors of the economy, and in almost all types of work, from managers (49% adoption) to personal service workers (12.5% adoption). So your babysitter or dogwalker may be using AI in their work, and in their lives, also.

Study shows AI adoption faster than PCs or the internet

OpenAI rolls out Advanced Voice assistant to all paid subscribers

OpenAI has been promising seamless voice interaction with its chatbots for a looong time - ever since their jaw-dropping demo last May. This week it finally released its Advanced Voice assistant, but only to paying customers. Early reviews are favorable, but there are some glitches. All the major AI players expect that voice assistants will be one of, if not the “killer app” that will unlock a flood of adoption by ordinary consumers. Apple is promising a smarter Siri “real soon now,” OpenAI has finally released its product to paying customers, while Google is fighting hard to make a dent in this space with its “Gemini Live” voice assistant which is available free to everyone with an Android phone.

OpenAI rolls out Advanced Voice assistant to all paid subscribers

AI2 releases Molmo, an open-source image analyzer

Paul Allen, billionaire co-founder of Microsoft, created the Allen Institute for Artificial Intelligence in 2014 as a nonprofit research institute. Previous projects have included an AI-enabled tracker of endangered species to prevent poaching, a system that monitors maritime traffic to prevent illegal fishing, and a system for climate modeling. This week they released Molmo, an open-source AI model that analyzes images. Its performance compares favorably with the large closed models like ChatGPT, even though Molmo is a fraction of the size, trained on a fraction of the data, and developed at a fraction of the cost of the closed models. More evidence that the creativity in the open-source community in AI is still a potent counterweight to the giant proprietary systems that tend to dominate the conversation.

AI2 releases Molmo, an open-source image analyzer

Try Molmo for free (no signup)

AI discovers 303 additional ancient images in Peruvian desert

Ancient civilizations in the Nazca desert of Peru carved huge figures, now known as “Nazca lines,” into the soil some 2000 years ago. The figures are so large that they can only be seen in their entirety from aerial photographs. Their purpose is still a mystery, although most scholars think they had ritual or religious significance. Since their rediscovery in the 1500s by Spanish conquistadores until last year, only some 430 of these figures had been discovered. Recently, scientists developed an AI model to scan satellite photographs of the Nazca Desert and identify possible additional figures. In 6 months of field work, the researchers confirmed 303 additional Nazca lines proposed by the AI system. They estimate that there are probably another 250 still-undiscovered figures, which they hope to find over the next few years by refining both the AI model and their field methods.

AI detects kids most vulnerable to RSV

Respiratory Syncytial Virus (RSV) can be deadly for the youngest patients, killing over 100,000 infants worldwide each year. Even in developed countries, the mortality rate is high, because the course of the illness is so unpredictable, and can transition from a mild illness to a life-threatening crisis within hours. Now two separate teams, one at Vanderbilt and another at Helsinki University, have each come up with AI models that help assess the risk of severe illness for each individual patient based on observable clinical parameters. Refinement of these systems should lead to better triage, and therefore better outcomes for these infants.

AI detects kids most vulnerable to RSV

Zzapp uses AI to fight malaria, at half the cost of bed nets

Nonprofit organization Zzapp Malaria uses an AI system to target mosquito-larvae elimination efforts in “hotspots” in countries where malaria is endemic. One trial in Sao Tome and Principe demonstrated a 74,9% reduction in mosquito population and a 52.5% reduction in malaria cases, at a cost of $0.44 per person protected, which is approximately half the cost of the current gold standard preventive treatment, bed nets soaked in insecticide. The AI system is used to direct the efforts of field workers who perform larvicidal interventions where the risk of transmission of malaria is greatest.

Zzapp uses AI to fight malaria, saving lives for less