The Download: AI transparency, and Twitter’s transformation

9 months ago 101

This is today's variation of The Download, our weekday newsletter that provides a regular dose of what's going connected successful the satellite of technology.

It’s precocious clip for much AI transparency

In little than a week since Meta launched its unfastened root AI model, LLaMA 2, startups and researchers person already utilized it to make a chatbot and an AI assistant. It volition beryllium lone a substance of clip until companies commencement launching products built with it.

LLaMA 2 makes a batch of sense. A nimble, transparent, and customizable exemplary that is escaped to usage could assistance companies make AI products and services faster than they could with a big, blase proprietary exemplary similar OpenAI’s GPT-4.

By allowing the wider AI assemblage to download the exemplary and tweak it, Meta could assistance to marque it safer and much efficient. And crucially, it could show the benefits of transparency implicit secrecy erstwhile it comes to the interior workings of AI models—at a constituent erstwhile that could not beryllium much timely, oregon much important. Read the afloat story.

—Melissa Heikkilä 

This communicative is from The Algorithm, her play AI newsletter. Sign up to person it successful your inbox each Monday.

If you’re funny successful learning much astir AI, cheque retired immoderate of our fantabulous caller reporting:

The must-reads

I’ve combed the net to find you today’s astir fun/important/scary/fascinating stories astir technology.

1 Twitter’s sanction alteration volition outgo its marque billions
‘Tweeting’ is portion of our taste lexicon—but Elon Musk doesn’t look to care.  (Bloomberg $)
+ He’s trying to destruct Twitter’s bequest and illustration a risky caller course. (WP $)

+ Musk is besides projecting ‘s3xy’ connected the company’s HQ walls. (NYT $)
+ If you’re a small spot bittersweet implicit Twitter’s demise, you’re not alone. (Vox)
+ The missive X does person an enduring benignant of appeal, though. (The Guardian)

2 The satellite isn’t prepared for what OpenAI’s been moving on
Sam Altman says his unit person created a unsafe AI they’ll ne'er release. (The Atlantic $)
+ His ambitions are progressively astatine likelihood with regulators. (FT $)
+ It’s clip to speech astir the existent AI risks. (MIT Technology Review)

3 Mastodon is rife with kid maltreatment images
It’s a grim reminder of however overmuch harder it is to mean decentralized platforms. (WP $)

4 China is fed up of Western sanctions
And it has nary qualms astir retaliating. (Economist $)
+ China is warring backmost successful the semiconductor exports war. (MIT Technology Review)

5 It’s harder for governments to emblem people’s societal media posts now
While the US has already been restricted, the aforesaid kinds of checks are besides coming to Europe. (Wired $)

6 America’s farmland is being turned into AI information centers
Sprawling centers are springing up successful agrarian expanses. (Insider $)

7 The FBI ran a concealed encrypted phones concern for years
Now, a squad of lawyers privation a justice to sanction the state that aided them. (Motherboard)
+ Erik Prince wants to merchantability you a “secure” smartphone that’s excessively bully to beryllium true. (MIT Technology Review)

8 Ozempic doesn’t person to outgo this much
But cause companies beryllium to marque money, aft all. (Slate $)
+ Weight-loss injections person taken implicit the internet. But what does this mean for radical IRL? (MIT Technology Review)

9 Influencers are emerging arsenic a almighty instrumentality for Indian politicians
Prime Minister Narendra Modi is seeking a 3rd word adjacent year, and doesn’t privation to instrumentality immoderate chances. (Rest of World)

10 We’re getting person to learning much astir asteroids ☄️
A ngo implicit a decennary successful the making is heading backmost to Earth—complete with immoderate precious stone and particulate cargo. (Ars Technica)

Quote of the day

"Is this what we want?"

—A elder decorator astatine Adobe questions the grade to which the institution should integrate AI into its products if it risks humans losing their jobs, Insider reports.

The large story

This institution is astir to turn caller organs successful a idiosyncratic for the archetypal time

August 2022

In the coming weeks, a unpaid successful Boston, Massachusetts, volition beryllium the archetypal to proceedings a caller attraction that could extremity up creating a 2nd liver successful their body. And that’s conscionable the beginning—in the months that follow, different volunteers volition trial doses that could permission them with up to six livers successful their bodies.

The institution down the treatment, LyGenesis, hopes to prevention radical with devastating liver diseases who are not eligible for transplants. Their attack is to inject liver cells from a donor into the lymph nodes of sick recipients, which tin springiness emergence to wholly caller miniature organs. 

These mini livers should assistance compensate for an existing diseased one. The attack appears to enactment successful mice, pigs, and dogs. Now we’ll find retired if it works successful people. Read the afloat story

Read Entire Article