Afrigather Plus: OpenAI could launch an online browser
That is at the moment’s version of The Obtain, our weekday publication that gives a day by day dose of what is going on on on the earth of know-how.
How OpenAI stress-tests its massive language fashions
OpenAI has lifted the lid (only a crack) on its safety-testing processes. It has put out two papers describing the way it stress-tests its highly effective massive language fashions to attempt to establish potential dangerous or in any other case undesirable habits, an method often known as red-teaming.
The primary paper describes how OpenAI directs an in depth community of human testers exterior the corporate to vet the habits of its fashions earlier than they’re launched. The second presents a brand new strategy to automate components of the testing course of, utilizing a big language mannequin like GPT-4 to give you novel methods to bypass its personal guardrails. MIT Expertise Assessment received an unique preview of the work.
—Will Douglas Heaven
Who ought to get a uterus transplant? Consultants aren’t positive.
Over 135 uterus transplants have been carried out globally within the final decade, ensuing within the births of over 50 wholesome infants. The surgical procedure has had profound penalties for these households—the recipients wouldn’t have been capable of expertise being pregnant some other means.
However authorized and moral questions proceed to encompass the process, which continues to be thought of experimental. Who ought to be provided a uterus transplant? Might the process ever be provided to transgender girls? And if that’s the case, who ought to pay for these surgical procedures? Learn the complete story.
—Jessica Hamzelou
This story is from The Checkup, our weekly publication concerning the newest in biotech and well being. Join to obtain it in your inbox each Thursday.
The must-reads
I’ve combed the web to seek out you at the moment’s most enjoyable/essential/scary/fascinating tales about know-how.
1 OpenAI could launch an online browser
Which might be a full-frontal assault on Google (The Data $)
+ The Google browser break-up is a solution in quest of a query. (FT $)
+ OpenAI by accident deleted potential proof in a coaching information lawsuit. (The Verge)
2 Border militias are prepared to assist with Trump’s deportation plans
No matter whether or not they’re requested to or not. (Wired $)
+ Trump’s administration plans to radically curb the powers of the federal company that protects unions. (WP $)
3 Russia hit Ukraine with a brand new kind of missile
Right here’s what we learn about it to date. (The Guardian)
4 Microsoft is about to show 50
And it’s each bit as related and highly effective because it’s ever been. (Wired $)
5 China has overtaken Germany in industrial robotic adoption
South Korea, nevertheless, stays streets forward of each of them. (Reuters $)
+ Three causes robots are about to develop into far more helpful. (MIT Expertise Assessment)
6 The irresistible rise of cozy tech
Our units, social media and now AI are encouraging us to maintain wanting inward. (New Yorker $)
+ Inside the comfortable however creepy world of VR sleep rooms. (MIT Expertise Assessment)
7 Churchgoers in a Swiss metropolis have been spilling their secrets and techniques to AI Jesus 😇
They usually’re largely actually having fun with it. Be careful, clergymen. (The Guardian)
8 A French startup desires to make gasoline out of skinny air
Then use it to gasoline ships and airplanes. (IEEE Spectrum)
+ All the things you’ll want to learn about various jet fuels. (MIT Expertise Assessment)
9 WhatsApp goes to start out transcribing voice messages
This appears compromise to bridge individuals’s completely different communication preferences. (The Verge)
10 Need a new telephone? It’s best to think about second-hand
It’s higher for the planet—and your pockets. (Vox)
Quote of the day
“Nope. 100% not true.”
—Jeff Bezos fires again at Elon Musk’s declare that he was telling everybody that Trump would lose pre-election in a uncommon publish on X.
The large story
This chemist is reimagining the invention of supplies utilizing AI and automation
DEREK SHAPTON
October 2021
Alán Aspuru-Guzik, a Mexico Metropolis–born, Toronto-based chemist, has devoted a lot of his life to considering worst-case eventualities. What if local weather change proceeds as anticipated, or will get considerably worse? Might we rapidly give you the supplies we’ll have to cheaply seize carbon, or make batteries from one thing aside from pricey lithium?
Supplies discovery—the science of making and creating helpful new substances—typically strikes at a frustratingly gradual tempo. The standard trial-and-error method takes a median of 20 years, making it too costly and dangerous for many corporations to pursue.
Aspuru-Guzik’s goal—which he shares with a rising variety of computer-savvy chemists—is to shrink that interval to a matter of months or years. And advances in AI, robotics, and computing are bringing new life to his imaginative and prescient. Learn the complete story.
—Simon Lewsen
We will nonetheless have good issues
A spot for consolation, enjoyable and distraction to brighten up your day. (Acquired any concepts? Drop me a line or tweet ’em at me.)
+ Do you wrestle with a insecurity? Right here’s how to take up a bit more room.
+ These recipes will guarantee you’ve got a scrumptious Thanksgiving subsequent week.
+ It’s not possible to not dream of lazy sunny days whereas gazing at Quentin Monge’s work.
+ Tom Jones x Disturbed = very humorous.
