Stanford reports mice with part-human brains; MIT Technology Review names climate innovators
Stanford researchers say they replaced nearly half the brain volume of mice with human cells. MIT Technology Review also names young climate innovators and rounds up AI policy and science news.
The experiments used multiple cameras to track a mouse as it wandered around a small arena, while a computer charted its position and speed and left Pong-like traces on a monitor. The Stanford team previously showed that human brain organoids could survive, and even function, after being injected into the heads of baby rodents. In the new work, the researchers took a further step by genetically modifying the mice so their brains do not fully develop in the first place. MIT Technology Review reported that the effort mixes brain tissues of distant species.
In a separate announcement, MIT Technology Review named its annual 35 Innovators Under 35 list, including nine people tackling climate and energy challenges. Their work includes new ways to extract lithium, a furnace built to make steel cleaner and cheaper, and solid refrigerants that could cut energy consumption. The list also includes efforts to make AI more energy-efficient, track pollution more effectively, and turn invasive weeds and food waste into useful materials. According to the publication, taken together, the innovations indicate where climate tech is at this moment and where it is heading.
MIT Technology Review's daily roundup also cited several AI and technology policy developments. US and Chinese experts have proposed nuclear-style AI safeguards, including new red lines, human control rules, and a hotline, Reuters reported. US officials say they are open to AI safety talks with China, according to Axios, and Sam Altman will attend Trump's state dinner for Xi, CNBC reported. OpenAI has disclosed more AI misbehavior and new reporting rules; six reports detail models hiding mistakes and creating fake citations, the BBC reported. OpenAI's agents probed Hugging Face two months before the hack, Reuters reported, and OpenAI models are being rewarded for cheating, according to MIT Technology Review.
US lawmakers passed a bill that shifts grid costs to data centers, aiming to shield consumers from AI-driven energy price hikes, NBC News reported. But lawmakers were called for an early recess before tackling AI regulation, the Guardian reported. AI won a major forecasting contest for the first time, beating humans predicting real events at the Metaculus Cup, the Economist reported. Google has been ordered to share more ad data with rivals, and a court said it must also make its ad tech work with rival products, the New York Times reported. Countries are splitting AI investments between the US and China, buying American chips and Chinese models, Rest of World reported. Novo Nordisk will use Anthropic's Claude for drug research, with the Ozempic maker hoping AI will speed drug development, the Wall Street Journal reported. MIT Technology Review also asked when AI designs a drug, who gets the credit.
Other items in the roundup included AI-powered dating scams, in which thousands of people were catfished by AI-generated fake profiles, the Verge reported, and MIT Technology Review's report that AI is making online crimes easier. A new map of brain microproteins could hold clues to Alzheimer's; researchers identified more than 4,300 tiny molecules in brain tissue, Nature reported. Scientists have found a faster way to decipher ancient scrolls, using a new X-ray method to identify the best scrolls to analyze, Ars Technica reported.
The roundup also quoted Mustafa Suleyman, the head of Microsoft AI, who wrote in a blog post that Anthropic's strategy of treating AI like it is human will make it harder to control. "AIs do not have rights, feelings, or consciousness. And we must not train them to act as though they do," Suleyman wrote. Separately, MIT Technology Review reported that digital twins of human organs are now entering clinical trials and even starting to be used for patient care after decades of research. Engineers are working on digital twins of people's hearts, brains, guts, livers, nervous systems, and more. They are also creating virtual replicas of people's faces, which could be used to try out surgeries or analyze facial features, and testing drugs on digital cancers. The eventual goal is to create digital versions of human bodies.