News: 1658160013

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

Cruise self-driving cars stopped and clogged up San Francisco for hours

(2022/07/18)


In brief Weeks after Cruise launched its autonomous, driverless taxi rides to the public in San Francisco, numerous vehicles mysteriously piled up and blocked several lanes of incoming traffic downtown.

Apparently a bunch of driverless Cruise vehicles were "stuck" blocking an intersection for "a couple of hours" the other night, according to a redditor who posted these images. No word on what the issue might have been. [1]https://t.co/EenuqbjTsD [2]pic.twitter.com/WkPQCF1SjL — E.W. Niedermeyer (@Tweetermeyer) [3]June 30, 2022

At least seven cars were spotted milling around a spot in San Franscico's Civic Center neighborhood at night. The driverless vehicles had stopped for some reason, preventing nearby traffic from moving. It's not clear why or how these cars seemingly suffered a technical glitch. Some of these issues were raised in an anonymous letter sent to the California Public Utilities Commission, who claimed Cruise is looking to launch its commercial robotaxi service too early, the Wall Street Journal [4]reported .

Tesla's head of AI leading Autopilot leaves

Andrej Karpathy, senior director of AI and an expert in computer vision helping Tesla develop self-driving cars, announced he was leaving after working at the company for five years.

It's been a great pleasure to help Tesla towards its goals over the last 5 years and a difficult decision to part ways. In that time, Autopilot graduated from lane keeping to city streets and I look forward to seeing the exceptionally strong Autopilot team continue that momentum. — Andrej Karpathy (@karpathy) [5]July 13, 2022

There were rumors Karpathy wasn't going to come back after he said he was taking a four-month sabbatical earlier this year in March, [6]according to Elektrek. Karpathy was hired to lead Tesla's AI and self-driving efforts in 2017 and left his previous role as a research scientist at OpenAI.

Top boss Elon Musk thanked him for his service via a message on [7]Twitter . Karpathy leaves at a dicey time for the company. Tesla's share price has dropped amid worsening market conditions, and it has shut down one of its offices in San Mateo. It is also facing heightened scrutiny that could lead to the National Highway Traffic Safety Administration [8]issuing a recall for hundreds of thousands of its cars.

Karpathy said he wasn't sure what he was going to do next but will focus on "technical work in AI, open source and education."

AI makes its way to the 2022 World Cup

AI-powered cameras will be deployed to help referees decide whether football players are offside in the upcoming 2022 World Cup tournament to be held in Qatar starting in November.

The technology involves placing a sensor inside the football and a series of cameras under the roof of the stadiums. The sensor will monitor its position on the football field, and the footage from cameras will be fed into machine learning algorithms capable of tracking the players' locations.

[9]People who regularly talk to AI chatbots often start to believe they're sentient, says CEO

[10]Cerebras sets record for 'largest AI model' on a single chip

[11]If AI chatbots are sentient, they can be squirrels, too

[12]Meta slammed with eight lawsuits claiming social media hurts kids

When the software detects a player is offside, an alert will be sent to people at a nearby control room. The information will be relayed to the referee, who will then decide whether to call the offence or not.

Pierluigi Collina, chairman of the FIFA Referees Committee, said the automated system will allow referees to make "faster and more accurate decisions," and said humans, not robots, were still in charge, [13]according to The Verge. Gianni Infantino, FIFA's current president, said the technology had been three years in the making and only takes seconds to call offside.

AI ethics review of research, yay or nay?

Academic conferences are asking AI researchers to consider in technical papers how their research could potentially lead to societal harm, and not everyone is happy.

As AI and machine learning technology continues to progress in academia, it's inevitable some of these techniques will end up being deployed in real life. The applications often reveal algorithms can be used for good and bad. Improved computer vision algorithms, for example, are helping develop self-driving cars but are also used for surveillance purposes.

[14]

AI-focused conferences like Neural Information Processing Systems and now the Conference on Computer Vision and Pattern Recognition are asking researchers to write paragraphs considering if and how their research could be harmful. But not everyone supports this initiative, Protocol [15]reported . Some researchers believe it's beyond the scope of their work or could impact research freedom, while others recognized their work could be abused in certain use cases.

[16]

[17]

"We're still at a point in AI ethics where it's very hard for us to properly assess and mitigate ethics issues without the partnership of folks who are intimately involved in developing this technology," said Alice Xiang, who leads Sony Group's AI ethics office and was a general co-chair at the ACM Conference on Fairness, Accountability, and Transparency.

Clearview fined by Greek authorities

Controversial facial recognition startup Clearview was fined €20 million for violating privacy laws by Greece's Hellenic Data Protection Authority (HDPA).

The company was accused of violating current EU GDPR rules by failing to obtain explicit consent to use individuals' personal data when it scraped billions of photographs posted on the internet. These images were used to build Clearview's database for its face-matching algorithms.

Given a picture, the company's software searches for potential matches against images in its database to reveal someone's identity by linking to their social media profiles, for example.

[18]

The fine was the largest amount ever ordered by the HDPA, a [19]according to The Record. A Clearview spokesperson claimed it "does not undertake any activities that would otherwise mean it is subject to the GDPR." ®

Get our [20]Tech Resources



[1] https://t.co/EenuqbjTsD

[2] https://t.co/WkPQCF1SjL

[3] https://twitter.com/Tweetermeyer/status/1542625065144946688?ref_src=twsrc%5Etfw

[4] https://www.wsj.com/articles/california-regulator-looking-into-anonymous-letter-alleging-cruises-robotaxi-service-wasnt-ready-for-launch-11657811102

[5] https://twitter.com/karpathy/status/1547332300186066944?ref_src=twsrc%5Etfw

[6] https://electrek.co/2022/07/13/tesla-loses-head-ai-vision-andrej-karpathy-big-loss-autopilot-team/

[7] https://twitter.com/elonmusk/status/1547332709025861632

[8] https://www.theregister.com/2022/06/09/nhtsa_tesla_autopilot/

[9] https://www.theregister.com/2022/07/04/ai_in_brief/

[10] https://www.theregister.com/2022/06/27/in_brief_ai/

[11] https://www.theregister.com/2022/06/20/in_brief_ai/

[12] https://www.theregister.com/2022/06/12/in-brief-ai/

[13] https://www.theverge.com/2022/7/5/23195009/fifa-ai-automated-offside-ball-body-tracking-2022-qatar-world-cup

[14] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2YtXYA2M9JRIhqPhHEQMFDAAAAMY&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0

[15] https://www.protocol.com/enterprise/ai-computer-vision-cvpr-ethics

[16] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44YtXYA2M9JRIhqPhHEQMFDAAAAMY&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[17] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33YtXYA2M9JRIhqPhHEQMFDAAAAMY&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0

[18] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44YtXYA2M9JRIhqPhHEQMFDAAAAMY&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[19] https://therecord.media/clearview-ai-fined-20-million-banned-from-processing-biometric-data-in-greece-after-gdpr-violations/

[20] https://whitepapers.theregister.com/



The robot uprising

Yet Another Anonymous coward

But instead of Schwarzenegger and the Terminator it's Reg from On The buses

Re: The robot uprising

bombastic bob

I was thinking more like "Johnny Cab" from 'Total Recall'

"I am not FAMILIAR with that address..."

Re: The robot uprising

Anonymous Coward

Hey Blakey does your bus go by the dark satanic mills?

Does it really need to be AI?

John Robson

Why not just put tags in the back of everyone's shirt and track that and the ball - no AI required...

If player's Z axis location is further forward than all but one of the opposition when the ball air pressure spikes (because it's being kicked) then flag potential offside - with an indicator on screen as to the player in question and the ball at moment of pressure spike.

Re: Does it really need to be AI?

Natalie Gritpants Jr

I'm confused. Does that mean you can't be taller than the opposition?

Re: Does it really need to be AI?

Yet Another Anonymous coward

Why not just put a series of long poles across the field

Introduce a large ball

Attach players to the poles and use large motors to spin the rods+players as the ball goes past

Re: Does it really need to be AI?

Falmari

@John Robson "ball air pressure spikes (because it's being kicked) then flag potential offside"

Needs to do more than just react to ball air pressure spikes otherwise the vast majority flagged as potential offside will not be offside.

The ball is kicked, headed, bounces, etc all the time, the majority of air pressure spikes are not due to a player releasing possession.

Also the system would have to identify which player and team was releasing possession to identify if that teams players were offside.

Back to the question Does it really need to be AI? No it does not.

Re: Does it really need to be AI?

Warm Braw

Or does it need to be done at all? I'm not really familiar with all the details of commercial lycra-based entertainment but I seem to recall they just toss a coin to decide who makes the first move. Couldn't they apply the same technology everywhere a decision is required?

Re: Does it really need to be AI?

Anonymous Coward

AI destroying more jobs, this time for the legally Blind.

Twenty million for scraping billions of images without permission?

Neil Barnes

Chump change, a cheap cost of doing business.

Make it illegal, full stop.

Re: Twenty million for scraping billions of images without permission?

Anonymous Coward

It *is* illegal, so there's the fine, and they must stop whatever they were doing amd delete it.

Since they continue asserting that their interpretation of the GDPR is the only one that matters, it might get interesting when they become a repeat offender.

Autonomous Vehicles all simultaneously updating their OS

bombastic bob

Windows Update strikes again

Deadlock condition?

Little Mouse

Those driverless Cruise cars sound like the church carpark when I was little, with all the drivers trying to our-nice each other when everyone left at the same time.

"After you." - "No no no. After you." - "But I insist. After you " etc etc etc.

Deadlock conditions are bad enough in software, without having them cross into the physical realm as well.

Re: Deadlock condition?

Yet Another Anonymous coward

Also rural Canada, people with right of way stop for other people at stop signs = chaos

Re: Deadlock condition?

Claptrap314

Sorry!

What matters...

Claptrap314

is intelligence of SOME form for detecting all the ********* **** ******* faked injuries.

DId I read this right

Anonymous Coward

Someone is really calling these things "Cruise" given their obvious use by terrorists as delivery systems?

Why bother with suicide bombers when you can just pack a taxis and send it to the desired address. Works for missiles too.

There was a mad scientist (a mad... social... scientist) who kidnapped
three colleagues, an engineer, a physicist, and a mathematician, and locked
each of them in separate cells with plenty of canned food and water but no
can opener.
A month later, returning, the mad scientist went to the engineer's
cell and found it long empty. The engineer had constructed a can opener from
pocket trash, used aluminum shavings and dried sugar to make an explosive,
and escaped.
The physicist had worked out the angle necessary to knock the lids
off the tin cans by throwing them against the wall. She was developing a good
pitching arm and a new quantum theory.
The mathematician had stacked the unopened cans into a surprising
solution to the kissing problem; his desiccated corpse was propped calmly
against a wall, and this was inscribed on the floor:
Theorem: If I can't open these cans, I'll die.
Proof: assume the opposite...