AI training AI and the future of robotaxis

AI training AI and the future of robotaxis

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

People who are paid to train AI are outsourcing their work to AI

The news: Many people who are paid to train AI models could outsource work for the AI, a new study has found.

This story is only available to subscribers.

Don’t settle for half the story.
Get paywall-free access to tech news for the here and now.

subscribe now
Already a subscriber? Registration

How they did: The researchers hired 44 people on the Amazon Mechanical Turk concert-working platform to summarize 16 excerpts from medical research papers, then analyzed their responses for telltale signs that they had been produced by the AI ​​models. They estimated that between 33% and 46% of workers used AI models like OpenAI’s ChatGPT to complete the task.

Because matter: Using AI-generated data to train AI could introduce further errors into already error-prone models. If they generate faulty output that is in turn used to train other AI models, the errors can be absorbed by those models and amplified over time, making their origins increasingly difficult to figure out. Read the full story.

Rhiannon Williams

To learn more about what happens when AI is trained on AI, check out my colleague Melissa Heikkils piece why we are hurtling towards a glitchy, spammy, fraudulent, AI-driven internet.

The robotaxis are here. It’s time to decide what to do about it

Benjamin Schneider

In some San Francisco neighborhoods, driverless cars have become a common sight. Many of the city’s spooky driverless cars are commercial robotaxis, directly competing with taxis and cab companies.

I’ve spent the last year covering robotaxis for the San Francisco Examiner, and have taken nearly a dozen driverless cruise car rides in the last few months. Throughout my reporting, I was struck by the lack of urgency in the public discourse around robotaxis. The time has come for the public and its elected representatives to play a more active role in shaping the future of this new technology. Like it or not, the robotaxis are here.

Now comes the difficult job of deciding what to do about it. Read the full story.

The chip-shaping machines that will shape computers will be next

When we talk about computer science these days, we tend to talk about software and the engineers who write it. But without the hardware and the physical sciences that made it possible—disciplines like optics, materials science, and mechanical engineering—modern computing would have been impossible.

Semiconductor lithography, the manufacturing process responsible for making computer chips, is at the center of a geopolitical race to control the future of computing power. And the rate at which new lithography systems and components are developed will determine not only the speed of computing progress, but also the balance of power and profit within the technology industry. Read the full story.

Chris Miller

Both Benjamin and Chris’ stories come from our next press issue, which is all about accessibility. If you haven’t already, subscribe to make sure you don’t miss out on future stories, subscriptions start at just $80 a year.

The wild ride to improve synthetic embryos

Antonio Regalado

Last week, The Guardian broke the news of a sensational breakthrough at the meeting of the International Society for Stem Cell Research in Boston.

The claim was that a researcher named Magdalena Zernicka-Goetz, based at the University of Cambridge and Caltech, had created synthetic human embryos using stem cells as a starting point.

What bothered me about this story is that this idea is not that new. The surprising fact that stem cells will self-organize into structures that share characteristics with real embryos has been known and studied for several years, as we first reported in 2017.

Since then, several labs have been in a competitive race to make these embryo models more complete, more realistic, and increasingly look like bona fide embryos, complete with placental tissue.

What annoyed the scientists was that Zernicka-Goetz appeared to claim that they had finished the race, but did so in the media, without providing any scientific proof.

The twist in the story is that there actually was a twist, but it came from a different lab Read the full story.

Antonio is replacing Jessica Hamzelou, who normally writes The Checkup, her weekly newsletter that gives you the inside track on all things biotech. Registration to receive it in your inbox every Thursday.

Required reading

I’ve scoured the internet to find you today’s funniest/important/scary/fascinating stories about technology.

1 The US Navy heard the Titan submersible implode days ago
But since it wasn’t definitive, officials decided to continue the search. (WSJ$)
+ The tragic mission now risks being consumed in messy legal proceedings. (Voice)
+ The operators CEO had an unconventional approach to security measures. (FT$)|
+ Director James Cameron says he’s heard about safety concerns about the submarine. (NYT$)

2 Generative AI models are not likely to comply with EU standards
Copyright, in particular, risks being a real sticking point. (FT$)
+ Some real evidence of why those AI stocks that steal our jobs are overrated. (wired $)
+ Stability AI wants the content of its models to appear even more realistic. (Bloomberg$)
+ Five big takeaways from the European AI law. (MIT Technology Review)

3 It has been a year since the US Supreme Court overturned Roe v. Wade
We still don’t fully understand what comes next. (WP$)
+ The cognitive dissonance of watching Roe’s demise unfold online. (MIT Technology Review)

4 Canada to force Google and Meta to pay for news
It’s possible the platforms could strike a deal, as Facebook did in Australia. (BBC)

5 Reddit is lobbying its moderators to end the blackout
Some subreddits are reopening under duress from the site. (The Edge)+ The company’s sales growth has stalled. (The information $)

6 The FBI demands police silence in exchange for surveillance help
The forces are forced to sign non-disclosure agreements before helping them track down the suspects. (wired $)

7 Electric bike fires are on the rise in New York
This year alone, 13 people have died, and gig economy workers are at high risk. (The Guardian)

8 The heatwaves are getting worse
And it is expected that they will become even more frequent. (Voice)
+ Do these heat waves mean that climate change is happening faster than expected? (MIT Technology Review)

9 How immunotherapy could help correct the immune system
Cancer treatment could treat the systems themselves, as well as attack rogue cells. (Economist $)+ Bacteria are also an effective way to treat some types of cancer. (New Scientist $)+ Research to show that biological sex matters in the immune system. (MIT Technology Review)

10 Liquid metal could turn any object into a smart device
Which could be a real help in space missions. (Ars Technique)

Quote of the day

I don’t have to trust or not trust.

Thierry Breton, Europe’s internal market commissioner, explains to a reporter that his recent trip to stress test Twitter’s compliance with Europe’s new digital content law is not based on his personal beliefs, reports The Wall Street Journal.

The big story

Inside the race to build the best quantum computer on Earth

February 2020

Google’s most advanced computer is not located at the company’s headquarters in Mountain View, California, or anywhere in the feverish expanse of Silicon Valley. It’s a few hours’ drive south in Santa Barbara, in a flat, soulless office park.

This is the computer Google is betting on to beat IBM in a race to be among the first to usher in a new era of machines that would make today’s most powerful computer look like an abacus, but through very different approaches. And it’s these different goals that could affect which of the two comes out ahead in the race for quantum computing. Read the full story.

Gideon Lichfield

We can still have nice things

A place of comfort, fun and distraction in these strange times. (Do you have any idea? Give me a call OR tweet it to me.)

+ A true Italian tried a deep-dish pizza and, of course, was not impressed.
+ How design collective Hipgnosis redefined album cover: from AC/DC to Pink Floyd.
+ Blogger-zines are back!
+ Remembering the incredible legacy of Alan Turing who was born on this day 111 years ago.
+ Drop everything: This is what it’s like to eat a $120,000 banana that’s also a work of art.


#training #future #robotaxis

Previous articleBlack Mirror’s big AI episode has the wrong villain
Next articleUnleash the power of Holoware computing with AMD Ryzen 7000 series processors

LEAVE A REPLY

Please enter your comment!
Please enter your name here