If two New Hampshire men aren't a match for the Devil, we might as well give the country back to the Indians. -Stephen Vincent Benet (1898-1943)
3D-printable concrete alternative hardens in three days, not four weeks (Ben Coxworth, November 25, 2025, new Atlas)
Although we’ve heard a lot about how 3D-printing concrete homes speeds up the construction process, you still have to wait up to 28 days for the concrete to sufficiently cure. A new printable substitute, however, is ready to go in just three days.
What’s next for AlphaFold: A conversation with a Google DeepMind Nobel laureate: “I’ll be shocked if we don’t see more and more LLM impact on science,” says John Jumper (Will Douglas Heaven, November 24, 2025, MIT Technology Review)
Proteins are made from strings of amino acids that chemical forces twist up into complex knots. An untwisted string gives few clues about the structure it will form. In theory, most proteins could take on an astronomical number of possible shapes. The task is to predict the correct one.
Jumper and his team built AlphaFold 2 using a type of neural network called a transformer, the same technology that underpins large language models. Transformers are very good at paying attention to specific parts of a larger puzzle.
But Jumper puts a lot of the success down to making a prototype model that they could test quickly. “We got a system that would give wrong answers at incredible speed,” he says. “That made it easy to start becoming very adventurous with the ideas you try.”
They stuffed the neural network with as much information about protein structures as they could, such as how proteins across certain species have evolved similar shapes. And it worked even better than they expected. “We were sure we had made a breakthrough,” says Jumper. “We were sure that this was an incredible advance in ideas.”What he hadn’t foreseen was that researchers would download his software and start using it straight away for so many different things. Normally, it’s the thing a few iterations down the line that has the real impact, once the kinks have been ironed out, he says: “I’ve been shocked at how responsibly scientists have used it, in terms of interpreting it, and using it in practice about as much as it should be trusted in my view, neither too much nor too little.” […]
AlphaFold was designed to be used for a range of purposes. Now multiple startups and university labs are building on its success to develop a new wave of tools more tailored to drug discovery. This year, a collaboration between MIT researchers and the AI drug company Recursion produced a model called Boltz-2, which predicts not only the structure of proteins but also how well potential drug molecules will bind to their target.
Last month, the startup Genesis Molecular AI released another structure prediction model called Pearl, which the firm claims is more accurate than AlphaFold 3 for certain queries that are important for drug development. Pearl is interactive, so that drug developers can feed any additional data they may have to the model to guide its predictions.
AlphaFold was a major leap, but there’s more to do, says Evan Feinberg, Genesis Molecular AI’s CEO: “We’re still fundamentally innovating, just with a better starting point than before.”
AI Is Suddenly Surprisingly Good At Physics (Sabine Hossenfelder, Nov 16, 2025)
LLMs aren’t able to actually use logic or reasoning to reach thought-out conclusions. Despite that, several startups plan on using the current systems to do serious physics research. And some physicists, including myself, have used AI chatbots like ChatGPT and Claude to write papers. The situation is changing incredibly fast. Let’s take a look at how LLMs might be improving at physics, and the current state of AI scientists.
A.I. Is a Bubble. Maybe That’s OK. (Mohamed A. El-Erian, 11/20/25, NY Times)
‘
But what if the bubble is an inevitable part of developing and adopting a revolutionary tool that will fundamentally improve productivity and growth? After all, A.I. is a general-purpose technology that will most likely alter a vast range of economic activities fundamentally. Its transformative potential could be on par with electricity, offering an enormous upside through durable improvements in what we do and how we do it. It’s not just that many existing activities will be done better and more efficiently. A.I. is poised to open the door to discoveries, particularly in health and education.
Such gains would allow the economy to grow faster without kicking off inflation, something economists describe as raising the “speed limit” for noninflationary growth. Increased productivity and a larger economy provide us with more opportunities to address the problems that my generation is leaving our kids and grandkids: high levels of debt, climate change and excessive income inequality.
Whichever way you look at it, the potential payoffs of A.I. adoption are staggering — for the economy, for social sectors, and, of course, for investors. That could not be said for the majority of the big historical bubbles, such as the tulip mania of the early 17th century.
AI in Medicine: Separating Silicon Valley Dreams from Scientific Reality (Mohammad FarhanNovember 16, 2025, Fair Observer))
During the COVID-19 pandemic, AI helped identify promising drug candidates and accelerated vaccine development timelines. Large language models are now scanning millions of research papers to identify potential therapeutic connections that would take human researchers years to discover.Meanwhile, in neuroscience, AI is being used to decode brain signals from paralyzed patients, enabling them to control computer cursors and robotic arms with unprecedented precision. Brain-computer interfaces powered by machine learning are translating neural activity into text, giving voice to patients who have lost the ability to speak. Researchers are using AI to map neural circuits with cellular precision and simulate brain networks that were previously too complex to model.
In structural biology, AI has achieved remarkable breakthroughs in protein structure prediction, which have major implications in drug discovery. Google DeepMind’s AlphaFold can now predict how proteins fold with stunning accuracy, solving a puzzle that has stumped scientists for decades. This matters because understanding protein structure is fundamental to developing new treatments for human diseases.
In drug discovery, we’re seeing real progress too. Companies like Exscientia made history with the molecule DSP-1181, the first AI-designed drug to enter human clinical trials for treating obsessive-compulsive disorder. In-silico Medicine became the first company to advance an AI-designed drug for an AI-discovered target into clinical trials — a “double first” where AI handled both target identification and drug design. Others, like Recursion Pharmaceuticals, have used AI to identify new drug targets and advance candidates like REC-1245 (an orally bioavailable molecular degrader of the RNA-binding protein 39) for solid tumors from discovery to pre-clinical testing in just 18 months, less than half the typical timeline.
Humans can no longer tell AI music from the real thing: survey (AFP, November 12, 2025)
The polling firm Ipsos asked 9,000 people to listen to two clips of AI-generated music and one of human-made music in a survey conducted for France-based streaming platform Deezer.
“Ninety-seven percent could not distinguish between music entirely generated by AI and human-created music,” said Deezer.
This AI Aced Hurricane Season in 2025. Here’s What That Means (Ellyn Lapointe, November 9, 2025, Gizmodo)
Though Google DeepMind’s Weather Lab only began releasing forecasts in June, it was by far the best model for predicting hurricane track and intensity this season, according to a preliminary analysis by Brian McNoldy, a meteorologist and senior researcher at the University of Miami. Meanwhile, America’s flagship weather model—the Global Forecast System—was the worst performing.
Coupert: smart shopping extensions are changing the way people save money online (Jon Stojan, November 3, 2025, Digital Journal)
The core advantage is automation. Instead of copying and pasting codes from coupon sites, smart tools like Coupert apply discounts instantly. Cashback rewards are layered in, and price history helps users identify better deals.
Honey: Applies automatic coupons but lacks integrated cashback features.
Rakuten: Provides strong cashback rates but requires manual activation and no coupon testing.
Capital One Shopping: Useful for price tracking, weaker in coupon coverage.
Ibotta: Well-known in groceries, but its receipt-based model is less efficient for online shopping.
RetailMeNot: Popular coupon site, but prone to expired and unverified codes.
Coupert integrates all of these features. Users save an average of $180.12 annually on coupons, $86.92 through cashback, and $333.74 via price tracking.
The end of the rip-off economy: From finance and medicine to used cars, artificial intelligence is radically improving market efficiency (The Econmist, 10/27/25)
IF YOU KNOW how to use artificial intelligence, it can save you a lot of time and money. Leasing a new car? Be sure to upload a photograph of the contract to ChatGPT first. Need help with a leaky tap? AI often understands the issue—and at a lower cost than a handyman. Parents with a fussy baby can now use chatbots to answer questions in seconds, rather than waiting for a doctor’s appointment. Giving Claude a PDF of a wine list is a great way to find the best-value bottles.
These examples add up to something bigger. As AI goes mainstream, it will remove one of the most enduring distortions in modern capitalism: the information advantages that sellers, service providers and intermediaries enjoy over consumers. When everyone has a genius in their pocket, they will be less vulnerable to mis-selling—benefiting them and improving overall economic efficiency. The “rip-off economy”, in which firms profit from opacity, confusion or inertia, is meeting its match.
Information advantages have existed for as long as markets themselves. In medieval England grocers used fake scales to dupe customers; pub landlords put salt in beer to make patrons thirstier. Such squalid practices are not just annoying. In a paper published in 1970, George Akerlof, a Nobel-prizewinning economist, discussed the market for used cars. It is hard for a buyer to know if such a car works properly or is a “lemon” with hidden problems. Buyers thus assume the worst. As a result, honest brokers, worried about being suspected of exploitative behaviour, stay away. The quality of service declines. Fewer consumers fulfil their needs.
The internet has made it harder to screw over customers. With Carfax and other providers of vehicle data, customers can check the history of a vehicle, overcoming some of the problems identified by Mr Akerlof. Taxi drivers now struggle to take people on circuitous but profitable routes, since apps such as Lyft and Uber tell them exactly where to go. Tripadvisor, a reviews website, sends tourists to restaurants that will provide a decent meal. In the early 2000s there were more than 20 branches of Angus and Aberdeen Steak Houses, a notorious tourist trap, in London. Today there are four, and the ones that remain are better than before.
Efficiency is deflationary.
AI Pessimism Fades as Reality Takes Hold (Brent Orrell, 10/30/25, AEIdeas)
In a large-scale Harvard Business School survey of 2,357 adults evaluating AI usage in 940 occupations, they found that reactions depend on how AI’s role is described. When AI was presented as a tool that augments rather than replaces human labor, a majority of Americans supported its use in 94 percent of occupations. Even when AI was described as fully automating core job tasks, respondents favored its use in 58 percent of occupations if it could outperform humans at a lower cost.
The key is specificity. When AI’s potential benefits are explained concretely—faster diagnoses, fewer repetitive tasks and injuries, better scheduling—attitudes shift from fear to interest. In the abstract, AI feels threatening; in context, it often looks like a gift.