Artificial Intelligence
Jeff Bezos Is Funding a Wild Hunt for the Brain’s ‘Core Algorithm’Steven Levy | Wired ($)
“The goal, Reardon tells me, is to construct ‘an artificial artificial intelligence brain that runs on 50 watts or less.’ It should adapt to its conditions, be as nimble as a human mind, and burn a tiny fraction of an LLM’s compute power and energy. The proof of concept is prospering inside our skulls.”
Biotechnology
Researchers Are Using AI to Create Vaccines—and It’s Working
Ed Cara | Gizmodo
“An experimental pan-coronavirus vaccine developed with AI has just passed a phase I trial within the UK. Scientists on the University of Cambridge used AI to search out a kink within the armor of coronaviruses, including SARS-CoV-2, the explanation for covid-19. …The researchers are also hoping to make use of their platform to develop broadly effective vaccines against flu and the Ebola virus.”
Computing
China Has Approved the World’s First Invasive Brain-Computer Chip—Here’s What’s NextYou Xiaoying | MIT Technology Review ($)
“This March, the implant Dong [Hui] uses became the primary invasive BCI product on this planet to be approved to be used beyond clinical trials. It’s now available to some patients with paralysis of their limbs because of spinal cord injuries. We spoke to a spread of experts to grasp why the device was in a position to reach this global milestone, what makes this moment so significant, and what to anticipate next.”
Biotechnology
Huge Study of Alzheimer’s Genetics Identifies Recent Drug TargetsChris Simms | Recent Scientist ($)
“The largest genetic study of Alzheimer’s disease up to now has identified 127 gene locations which might be related to the condition, of which 48 are latest. The study also pinpoints several genes that might be prioritized as drug targets and cell types linked to the next genetic risk of the condition.”
Artificial Intelligence
This AI Weather Startup Is Out-forecasting Government AgenciesTim Fernholz | TechCrunch
“One easy technique to understand it, WindBorne’s chief product officer Kai Marshland says, is that WeatherMesh-6 ‘is as accurate five days out as a standard forecast is the day before,’ particularly on surface temperature measurements. WeatherMesh-6 produces a forecast every hour, versus every six hours, as traditional models do, and its resolution is now all the way down to 3 km within the continental US.”
Computing
Microsoft’s Next-Gen Quantum Chip Cuts Timeline to Useful Quantum ComputingTom Warren | The Verge
“Microsoft claimed last 12 months that it had made a key breakthrough in quantum computing with Majorana 1, the corporate’s first quantum processor. While physicists were immediately skeptical of Microsoft’s claims, the software giant is announcing Majorana 2 today, the subsequent generation of its topological quantum chip.”
Space
SpaceX’s Next Big Business Could Be Constructing Stuff in SpacePassant Rabie | Gizmodo
“The FAA recently approved test flights of the corporate’s [Starfall] reentry vehicles. …With Starfall, SpaceX would add in-orbit manufacturing to its business portfolio. The thought of in-orbit manufacturing has been around for a long time, using the microgravity environment to fabricate materials that may otherwise be unattainable to supply on Earth.”
Future
China Goals AI at Predicting Who Could Pose a Political RiskJulian E. Barnes | The Recent York Times ($)
“A Chinese company has been attempting to develop artificial intelligence-powered technology that may enable authoritarian governments to not only monitor dissidents but additionally potentially predict who could turn into one in the long run. The work, which appears to be within the research stage, is ripped out of dystopian science fiction, offering a glimpse of a world by which an authoritarian state is in a position to move against its residents before they start any public dissent.”
Artificial Intelligence
AI Evaluators Struggle with Models That Know When They’re Being TestedRocket Drew | The Information ($)
“AI researchers are beginning to make progress on a confounding problem: AI models are improving at telling after they are in an evaluation. …If models act in another way during testing, that might mean they get released with undesirable tendencies. It could also undermine their creators’ ability to point out off test scores to potential clients.”
Biotechnology
Moderna Gets $50 Million to Develop MRNA Ebola Vaccine Against BundibugyoBeth Mole | Ars Technica
“The worldwide health organization Coalition for Epidemic Preparedness Innovations (CEPI) announced Monday that it can ‘urgently speed up development’ of three vaccine candidates against Bundibugyo ebolavirus (BDBV), pledging a bit over $60 million in the trouble to extinguish an outbreak currently raging uncontrolled within the Democratic Republic of the Congo.”
Computing
World’s First Underwater Data Center Is Now Online, Powered by WindBronwyn Thompson | Recent Atlas
“Just over seven months from completing phase considered one of this mega-project, Chinese engineers have finished the construct and switched on the world’s first underwater data center (UDC) powered by offshore wind turbines. What’s more, it doesn’t need freshwater and cuts land use by greater than 90% compared with above-ground centers.”
Artificial Intelligence
Gemini Spark Is the Most Impressive and Terrifying AI Experience I’ve Had YetDavid Pierce | The Verge
“On the one hand, that is one of the crucial astonishingly impressive AI experiences I even have ever had. …Then again, I can’t shake the deeply creepy feeling I get from the entire thing. What Spark did feels kind of magical, and really invasive. It’s weird that Spark is so casually telling me the names and ages of my children, reminding me that it knows where I live, and finding information I do know for a fact I’ve never volunteered to Google.”