The Obtain: Russia’s crumbling tech trade, and an AI safety catastrophe

That is at present’s version of The Obtain, our weekday e-newsletter that gives a each day dose of what’s happening on the planet of know-how.

How Russia killed its tech trade

Within the months after Vladimir Putin introduced the invasion of Ukraine, Russia noticed a mass exodus of IT employees. In line with authorities figures, about 100,000 IT specialists left Russia in 2022, or some 10% of the tech workforce—a quantity that’s doubtless an underestimate.

It has now been over a yr because the invasion started. The tech employees who left all the things behind to flee Russia warn that the nation is nicely on its solution to changing into a village: minimize off from the worldwide tech trade, analysis, funding, scientific exchanges, and important elements. It’s an accelerating development that began nicely earlier than the struggle. Learn the complete story.

—Masha Borak

3 ways AI chatbots are a safety catastrophe 

AI language fashions are the shiniest, most fun factor in tech proper now. However they’re poised to create a significant new drawback: they’re ridiculously straightforward to misuse. No programming expertise are wanted, and there’s no identified repair.

Tech corporations are racing to embed these fashions into tons of merchandise to assist individuals do all the things from e book journeys to arrange their calendars to take notes in conferences. 

However the best way these merchandise work creates a ton of latest dangers, from leaking individuals’s personal info to serving to criminals phish, spam, and rip-off individuals. Our senior AI reporter Melissa Heikkilä has dug into the methods they’re open to abuse. Learn the complete story.

When you’d prefer to learn extra in regards to the safety vulnerabilities lurking in AI merchandise, Melissa has written about why we’re hurtling towards a glitchy, spammy, scammy, AI-powered web for The Algorithm, her weekly e-newsletter. Join to obtain it in your inbox each Monday.

The complicated math of counterfactuals may assist Spotify decide your subsequent favourite track

The information: A brand new type of machine-learning mannequin constructed by a workforce of researchers on the music-streaming agency Spotify captures, for the primary time, the complicated math behind counterfactual evaluation, a way that can be utilized to establish the causes of previous occasions and predict the results of future ones.

What are counterfactuals? The fundamental concept behind counterfactuals is to ask what would have occurred in a state of affairs had sure issues been totally different. It’s like rewinding the world, altering just a few essential particulars, after which hitting play to see what occurs. By tweaking the correct issues, it’s potential to separate true causation from correlation and coincidence.

Why it’s vital: The mannequin may enhance the accuracy of automated choice making, particularly customized suggestions, in a spread of purposes past track recommendations, from finance to healthcare. Learn the complete story.

—Will Douglas Heaven

The must-reads

I’ve combed the web to search out you at present’s most enjoyable/vital/scary/fascinating tales about know-how.

1 NASA has named the astronauts that can head again to the moon
It’ll be the primary crewed moon mission since Apollo in 1972. (BBC)
+ The mission’s crew will fly to the moon subsequent yr. (The Atlantic $)
+ Richard Branson’s Virgin Orbit has filed for chapter. (Sky Information)
+ South Korea’s Hanwha is hoping to take its place as a SpaceX challenger. (Bloomberg $)
+ What’s subsequent in area. (MIT Expertise Overview)

2 ICE is demanding information from colleges and abortion clinics
It’s a technique that might be unlawful. (Wired $)
+ Texas is attempting out new ways to limit entry to abortion drugs on-line. (MIT Expertise Overview)

Three AI textual content detectors aren’t working
A instrument designed to flag AI textual content penalized an harmless highschool pupil as a substitute. (WP $)
+ Universities aren’t satisfied by the software program’s guarantees. (FT $)
+ How OpenAI snowballed from a plucky startup to an AI big. (The Data $)
+ Why detecting AI-generated textual content is so tough (and what to do about it) (MIT Expertise Overview)

four Paris needs its flying taxis up and working by subsequent yr’s Olympics
It’s an optimistic intention, to place it mildly. (Bloomberg $)
+ These plane may change how we fly. (MIT Expertise Overview)

5 Ozone-depleting chemical substances are making a comeback
Scientists are struggling to work out what’s inflicting the rise in emissions. (The Verge)
+ The chemical substances have been banned since 2010. (New Scientist $)

6 The lure of chatbots for political pollsters
People don’t reply the cellphone, however chatbots are all the time prepared to speak. (The Atlantic $)

7 Australia has banned TikTok on authorities gadgets
It’s the most recent in a protracted line of nations erring on the aspect of warning. (TechCrunch)
+ The fantastic thing about TikTok’s secret, stunning, and eerily correct suggestion algorithms. (MIT Expertise Overview)

eight Dad and mom are reserving social media handles for his or her infants 🍼
However there’s no assure they’ll really need to use them. (NYT $) 

9 AI doesn’t have a way of scent
But it surely’s getting used to design bespoke fragrances anyway. (FT $)

10 You possibly can’t escape voice notes 🔊
They’re low effort for the sender, however could be a trouble for the receiver. (Vox)

Quote of the day

“It’s the subsequent step on the journey that will get humanity to Mars. This crew will always remember that.”

—Victor Glover, one of many astronauts because of journey to the moon, describes the  significance of NASA’s Artemis II mission, Ars Technica reviews.

The large story

What does GPT-3 “know” about me? 

August 2022

One of many greatest tales in tech is the rise of enormous language fashions that produce textual content a human may need written. 

These fashions’ energy comes from troves of publicly accessible human-created textual content that has been hoovered from the web. When you’ve posted something even remotely private in English on the web, chances are high your information is perhaps a part of a number of the world’s hottest LLMs. 

Melissa Heikkilä, MIT Expertise Overview’s AI reporter, puzzled what information these fashions may need on her—and the way it may very well be misused. So she put OpenAI’s GPT-Three to the check. Examine what she discovered.

We are able to nonetheless have good issues

A spot for consolation, enjoyable and distraction in these bizarre instances. (Obtained any concepts? Drop me a line or tweet ’em at me.)

+ How DJ Khaled spearheaded rap’s love of memes. 
+ This lady’s work: these photographs of girls by the ages are fairly one thing.
+ The science behind how and why we fall in love is fascinating.
+ The heartwarming story behind the creation of Tetris.
+ What do faux pop stars inform us about the actual world? Quite a bit, really.

Leave a Reply

Your email address will not be published. Required fields are marked *