The Download: how to test AI, and the hidden victims of pig-butchering scams


via MIT Technology Review https://ift.tt/eRmbwpE

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

Large language models aren’t people. Let’s stop testing them as if they are.

In the past few years, multiple researchers claim to have shown that large language models can pass cognitive tests designed for humans, from working through problems step by step, to guessing what other people are thinking. 

These kinds of results are feeding a hype machine predicting that these machines will soon come for white-collar jobs; that they could replace teachers, doctors, journalists, and lawyers. Geoffrey Hinton has called out GPT-4’s apparent ability to string together thoughts as one reason he is now scared of the technology he helped create.

But there’s a problem. There’s little agreement on what those results really mean. Some people are dazzled by what they see as glimmers of human-like intelligence, while others aren’t convinced one bit. And the desire to anthropomorphize such models is confusing people about what they can and cannot do. Read the full story.

—William Douglas Heaven

The involuntary criminals behind pig-butchering scams

Pig-butchering scams are everywhere. The scams, the term for which refers to the lengthy, trust-building process of raising a pig for slaughter, have extorted victims out of millions, if not billions, of dollars.

But in recent weeks, growing attention has been granted to the scammers behind these crimes, who are often victims themselves. A new book in English, a movie in Chinese, and a slew of media reports are shining a light on the fascinating (and horrifying) aspects of a scary trend in human trafficking, where victims leave their homes in the hope of gaining stable employment, but end up held captive and unable to leave. Read the full story.

—Zeyi Yang

Zeyi’s story first appeared in China Report, his weekly newsletter giving you the inside track on all things tech in China. Sign up to receive it in your inbox every Tuesday.

The must-reads

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

1 The FBI has dismantled a colossal malware botnet
It had infected more than 700,000 computers across the world. (The Verge)
+ It’s the most sophisticated botnet the authorities have ever encountered. (The Register)
+ Russian cybercrime forums are offering big cash prizes for scam tutorials. (Wired $)

2 Big Tech is propping up deepfake porn
Its hosting infrastructure spreads non-consensual material to wide audiences. (Bloomberg $)
+ A horrifying AI app swaps women into porn videos with a click. (MIT Technology Review)

3 Google has unveiled a suite of new corporate AI tools
The goal is to put its AI-powered office software in the hands of as many customers as possible. (WSJ $)

4 Amazon is facing legal action over the sales of unapproved drugs
It’s been warned multiple times in the past year to stop selling unproven medicines. (FT $)

5 A new nuclear arms race is beckoning
Relations between the US, Russia, and increasingly China, are growing increasingly fraught. (Economist $)

6 Arizona’s chip factory is struggling to get online
America’s home-built chip ambitions are off to a rocky start. (The Guardian)
+ The $100 billion bet that a postindustrial US city can reinvent itself as a high-tech hub. (MIT Technology Review)

7 Alternative meats are hot right now 🍗
Governments have sunk $1 billion into no-kill meat. But will we eat it? (Vox)
+ Will lab-grown meat reach our plates? (MIT Technology Review)

8 How to talk to whales—using AI 🐋
Scientists are deciphering the mammals’ speech patterns, and creating their own along the way. (Wired $)
+ Under-represented languages pose a major obstacle for AI models. (Rest of World)

9 These materials are being used to build the cities of the future
Greener wood, cement, and glass is on the horizon. (Bloomberg $)
+ How green steel could clean up a dirty industry. (MIT Technology Review)

10 Robotaxis are weird now
Just ask the people who live in the cities they’re clogging up. (The Atlantic $)
+ Robotaxis are here. It’s time to decide what to do about them. (MIT Technology Review)

Quote of the day

“The lack of crouching is probably the most glaring issue.”

—Virtual reality enthusiast Brad Lynch offers his opinion on Meta’s decision to finally add legs to its VR avatars.

The big story

We asked Bill Gates, a Nobel laureate, and others to name the most effective way to combat climate change

 

February 2021

Despite decades of warnings and increasingly devastating disasters, we’ve made little progress in slowing climate change.

Given the lack of momentum, how do we make faster, more significant progress? We asked 10 experts a single question: “If you could invent, invest in, or implement one thing that you believe would do the most to reduce the risks of climate change, what would it be and why?” Read the full story.

—James Temple

We can still have nice things

A place for comfort, fun and distraction in these weird times. (Got any ideas? Drop me a line or tweet ’em at me.)

+ If you’re lucky enough to see the super blue moon this week, here’s how to maximize your chances of taking the best photo possible.
+ What did druids really get up to?
+ McDonald’s McFlurry machines are always broken just when you need them. Here’s why.
+ The making of Jamiroquai’s Virtual Insanity music video is actually insanely cool.
+ Congratulations to Python, which has been named this year’s top programming language.

Comments