the workers who train the AI system, don’t trust the technology to drive them.
The report, based on interviews with nine former Tesla data labelers, a former self-driving engineer, and 11 traffic-safety researchers, paints a damning picture of the gap between Tesla’s safety marketing and the reality of its autonomous driving program.
Reuters found that a central comparison error inflated Tesla’s claimed safety level by a factor of three. Tesla counted crashes where airbags deployed in its own vehicles, then compared that number to federal data that includes all crashes requiring a tow truck — a far less severe threshold. Tow-truck crashes often don’t involve airbag deployments at all.
The critical point: the federal data Tesla used already included airbag-deployment crashes as a separate category. Tesla could have made a valid apples-to-apples comparison but chose not to.
When University of Michigan researcher Marco Benedetti performed the correct comparison — airbag crashes for Teslas versus airbag crashes for all vehicles — the result dropped from “10 times safer” to roughly three times farther between crashes. And even that figure is unreliable because of additional methodological problems, including the massive age gap between Tesla’s fleet (4.1 years average) and the overall U.S. fleet (12.8 years).
What I get from the article.
Former Tesla employees-perhaps disgruntled former employees?
It seems to me that Tesla used the wrong comparison to inflate the safety statistic. Someone, somewhere in the company made that decision.
3 times safer is not 10 times safer. But it is still a good number.
My thoughts about Tesla EV business. Tesla makes quality well made EVs that used to be the standard other EVs are measured against. IMO that no longer is the case. Some Chinese EV manufacturers are turning out superior EV product. But you will not find that in their $15-25k EV models, but in $40-45k & up EVs.
And I don’t believe any Tesla statement about FSD. I am from Missouri on that. Show ME.
Thing is, Tesla doesn’t need to fabricate their stats, or otherwise hype their ADAS. It works great, one of the best in the industry.
They are starting to run into trouble because they can’t keep the hype balloon going forever. It is becoming increasingly clear that Tesla can’t provide L4 autonomy with HW4. So they do things like announce a robotaxi service—but with safety drivers. The announce the Cybercab will be in mass production in April–but they don’t have any place to deploy it.
Problem is you can only make those announcements once, then you need a new announcement.
Now could be the right timing, to help us forget about autonomous cars and humanoid robots and focus on the much larger rest of the universe with its natural synergies between AI and space.
From Texas regulators,
Tesla now has 42 autonomous vehicles authorized for driverless ridehailing in Texas, a fleet that’s less than one-tenth the size of Waymo’s in the state.
(note that authorized says nothing about performance or safety or availability or much else)
There are no “natural synergies” of AI and space. Energy, as on earth, is a constraint. Solar panels need to be large, which induces atmospheric drag, which shortens life. Heat, as on earth, is a constraint. Radiation into a vacuum is grossly inefficient (thermos bottle, tip o’ the hat to McLovin’) meaning panels will be large. Chip life is noted to be about 5 years, making failure and frequent replacement necessary. This is done more easily on the ground than in a satellite whizzing past.
This is not to say it can’t be done. Many things can be done, and are, but that doesn’t mean they make sense.
Now a lot of this could change, if more efficient chips or some substitute is made which 1) uses less energy and 2) produces less heat, (an both those issues are improving) but we ar a long way from the kind of numbers that make this much more than blunt science fiction for now.
Most of your arguments make a lot of sense. But chips can last a lot longer than 5 years if they are operated such that they don’t reach extreme temps (i.e. running them at very high clock speeds). They can just run them at a bit lower speeds and get them to last longer in space. They can also have software that self checks each chip and disables a chip if it is failing, keeping the rest of them on that satellite on line. This still doesn’t make data centers in space a good idea.