The Download: the mystery of LLMs, and the EU’s Big Tech crackdown

This is today's edition ofThe Download,our weekday newsletter that provides a daily dose of what's going on in the world of technology.

Large language models can do jaw-dropping things. But nobody knows exactly why.

Two years ago, Yuri Burda and Harri Edwards, researchers at OpenAI, were trying to find out what it would take to get a large language model to do basic arithmetic. At first, things didn't go too well. The models memorized the sums they saw but failed to solve new ones.

By accident, Burda and Edwards left some of their experiments running for days rather than hours. The models were shown the example sums over and over again, and eventually they learned to add two numbers-it had just taken a lot more time than anybody thought it should.

In certain cases, models could seemingly fail to learn a task and then all of a sudden just get it, as if a lightbulb had switched on, a behavior the researchers called grokking. Grokking is just one of several odd phenomena that have AI researchers scratching their heads. The largest models, and large language models in particular, seem to behave in ways textbook math says they shouldn't.

This highlights a remarkable fact about deep learning, the fundamental technology behind today's AI boom: for all its runaway success, nobody knows exactly how-or why-it works. Read the full story.

-Will Douglas Heaven

If you're interested in the mysteries of AI, why not check out:

+ Why AI being good at math matters so much-and what it means for the future of the technology.

+ What the history of AI tells us about its future. IBM's chess-playing supercomputer Deep Blue was eclipsed by the neural-net revolution. Now, the machine may get the last laugh. Read the full story.

+ What an octopus's mind can teach us about AI's ultimate mystery. Machine consciousness has been debated-and dismissed-since Turing. Yet it still shapes our thinking about AI. Read the full story.

+ The great AI consciousness conundrum. Philosophers, cognitive scientists, and engineers are grappling with what it would take for AI to become conscious. Read the full story.

The must-reads

I've combed the internet to find you today's most fun/important/scary/fascinating stories about technology.

1 The EU wants to end self-regulation for Big Tech
A new law is forcing the biggest companies to alter their technologies-or else. (NYT $)
+ The EU wants to regulate your favorite AI tools. (MIT Technology Review)

2 The US-China trade secret war is escalating
Proving criminal intent is tough, but the risks posed by corporate espionage are real. (WSJ $)
+ The Chinese Communist Party is gearing up for its annual meeting in Beijing. (Economist $)
+ Meanwhile, China is offering AI startups subsidizing vouchers. (FT $)

3 Apple's driverless car project was always doomed
The company has little to show for nearly a decade of work on it. (The Verge)
+ Its hopes of dethroning Tesla were ridiculously ambitious. (Bloomberg $)
+ Apple is no stranger to axing a project it feels isn't working. (TechCrunch)

4 Brace yourself for generative AI's data-stealing worms
It's an entirely new kind of cyberattack. (Wired $)
+ India is cracking down on unreliable and unfinished AI tools. (Reuters)
+ Three ways AI chatbots are a security disaster. (MIT Technology Review)

5 Lab testing monkeys are being laundered from Cambodia
They're ripped from their homes, illegally bred, and shipped into the US. (Bloomberg $)

6 We're being advised to treat covid like the flu
New US guidance says we don't need to isolate anymore. (NY Mag $)
+ Instead, stay at home until you feel well enough to mix again. (Ars Technica)
+ Scientists are finding signals of long covid in blood. (MIT Technology Review)

7 Waymo's driverless cars are heading to LA
Despite some very vocal backlash from locals. (Insider $)
+ What's next for robotaxis in 2024. (MIT Technology Review)

8 It's high time we killed our speeding habit
Modern vehicles are capable of unnecessarily high speeds. A new technology could stop that. (Vox)

9 Meet the AI influencers changing the course of the industry
It goes to show X is still good for something. (IEEE Spectrum)
+ Why it matters how we picture AI. (New Yorker $)
+ I went for a walk with Gary Marcus, AI's loudest critic. (MIT Technology Review)

10 We're getting closer to finding real-world use cases for quantum computers
And Google's willing to pay to hear your ideas. (New Scientist $)
+ How scientists are using quantum squeezing to push the limits of their sensors. (MIT Technology Review)

Quote of the day

To be honest, what concerns me more is the lack of control from humans, who are still making wars after 2,000 years."

-Roberto Cingolani, CEO of Italian defense company Leonardo, explains to CNBC why he thinks stupidity is a greater threat to humanity than AI.

The big story

Humanity is stuck in short-term thinking. Here's how we escape.

October 2020

Humans have evolved over millennia to grasp an ever-expanding sense of time. We have minds capable of imagining a future far off into the distance. Yet while we may have this ability, it is rarely deployed in daily life. If our descendants were to diagnose the ills of 21st-century civilization, they would observe a dangerous short-termism: a collective failure to escape the present moment and look further ahead.

The world is saturated in information, and standards of living have never been higher, but so often it's a struggle to see beyond the next news cycle, political term, or business quarter. How to explain this contradiction? Why have we come to be so stuck in the now"? Read the full story.

-Richard Fisher

We can still have nice things

A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or tweet 'em at me.)

+ Uhoh: maybe you can end a sentence with a preposition after all?
+ This canoeist did not have to make crossing a river this arduous.
+ Ever wondered how the legendary lyrics to Gorillaz's Dare came about? Wonder no more!
+ Justice for The Phantom Menace (no really).
+ Step away from the purring cat (NYT $)