Elon: FSD Beta tweets

[EVNow]

Again, this is most definitely NOT known! We do not have the data. It has not been provided.

The data has been provided - but you are disputing it. Which is fine … but let’s be clear about the facts.

Specific example:
Personally, if I were in a situation where I felt an accident was more likely, or a particularly complex driving situation, I would often not be using FSD.

Possible - but then so many other things. Your example here can’t be disproven - so not very relevant.

Ofcourse the academic way of doing this would be to provide all the data and publish a paper in a journal so others can see all the data and check all assumptions etc. We are not getting that kind of transparency from a private company anytime soon. May be when they need to get approval from EU for actual FSD in a decade - but by then this discussion would be moot.

 

[AlanSubie4Life]

The data has been provided - but you are disputing it.

Please show me where the data show that FSD (I mean FSD Beta City Streets of course - that is the context here) + human is safer than human alone. There is no clear control group to compare to (note that Telsa can likely produce such a control dataset quite easily)!

I am absolutely NOT disputing any of the data provided by Tesla. Those data are likely rock solid. The methodology seems sound and it’s not sampled data.

example here can’t be disproven - so not very relevant.

It was just an example. A thought experiment. It isn’t necessary to prove or disprove it and it may not be widely applicable. We just don’t know.

 

[Bladerskb]

I don't know what about my statement you think is not provable with the data provided.

There are 0.31 accidents per million miles where FSD Beta is engaged; whether you want to dispute the accuracy of that fact is separate from the concept that it's a low level of accidents and objectively safe. This figure of 0.31 accidents per mile takes into account all the times that humans intervened and prevented an accident. So it doesn't matter if hypothetically FSD Beta is 10x less safe than manual driving, the end result of the entire Beta program is objectively safe.

What data? I see no data provided. I see a sentence, that's no data.

The data has been provided - but you are disputing it. Which is fine … but let’s be clear about the facts.

What DATA? I would love to see the DATA! Show me the data you are referring to?

 

[AlanSubie4Life]

What data? I see no data provided. I see a sentence, that's no data.

There are several per-million-mile numbers (or the inverse) presented which are relevant. As I have repeatedly said, they are quite lacking in detail of exactly what they cover (what does “Mostly Non-Highway Miles” mean, for example?).

And then separately there are the Autopilot/Non-AP rates, but no mention of the total driven miles for each dataset so it makes it difficult to calculate overall rates and make comparisons.

There are definitely a lot of pieces of crucial information missing, but there is definitely data, providing summary numbers from a very robust dataset.

 

[Dewg]

Tesla Full Self-Driving (FSD) Beta program surpasses 150 million miles

Tesla's FSD Beta program has reached a significant milestone, attaining over 150 million miles driven to date.

www.teslarati.com

Tesla FSD Milestone: FSD Now Drives 1 Million Miles Per Day

Tesla's latest Full Self-Driving milestone, as the fleet achieves 1 million miles driven per day, with rapid improvements and simulated training propelling auto

www.notateslaapp.com

Looks like we're in pretty big numbers for miles driven on Beta. Autopilot numbers are in the single digit billions of miles (last estimate was over 5 billion in 2021).

 

[DarkandStormy]

Elon Musk’s statements could be ‘deepfakes’, Tesla defence lawyers tell court

Judge in Autopilot death case says defence argument ‘deeply troubling’ and wants Tesla CEO interviewed under oath on safety claims

www.theguardian.com

So are all his tweets also deep fakes?

 

[diplomat33]

Elon Musk’s statements could be ‘deepfakes’, Tesla defence lawyers tell court

Judge in Autopilot death case says defence argument ‘deeply troubling’ and wants Tesla CEO interviewed under oath on safety claims

www.theguardian.com

So are all his tweets also deep fakes?

And what about the live presentations like Autonomy Day, AI Day and shareholder meetings? Elon was in-person on stage when he made comments about FSD. Those cannot be deepfakes. This Tesla legal defense is truly absurd and bizarre.

 

[DarkandStormy]

And what about the live presentations like Autonomy Day, AI Day and shareholder meetings? Elon was in-person on stage when he made comments about FSD. Those cannot be deepfakes. This Tesla legal defense is truly absurd and bizarre.

My guess is they're feeling the heat from lawsuits piling up, so they have to throw out something as a defense. But I'm not a lawyer.

 

[diplomat33]

My guess is they're feeling the heat from lawsuits piling up, so they have to throw out something as a defense. But I'm not a lawyer.

Definitely. The "deepfakes" defense seems like a desperate move to do something because they have no real defense. Elon's statements are public domain and clear as day. They can't refute them.

 

[Goose66]

And what about the live presentations like Autonomy Day, AI Day and shareholder meetings? Elon was in-person on stage when he made comments about FSD. Those cannot be deepfakes. This Tesla legal defense is truly absurd and bizarre.

As was the case here, where the statements the Plaintiff are relying on were made in public at a 2016 Code Conference. There's 500 other people that could be deposed right there. This is sheer desperation on the part of the defense counsel and I think strips them of all credibility.

 

[willow_hiller]

Definitely. The "deepfakes" defense seems like a desperate move to do something because they have no real defense. Elon's statements are public domain and clear as day. They can't refute them.

It seems like a pretty straightforward defence to me, being blown out of proportion.

They are suing claiming Elon's statements led the deceased to believe he could play games on his phone and not pay attention to the road. Elon made no such statements about the abilities of Autopilot at the time of the crash. Therefore if there are any recorded statements of Elon saying "You don't need to pay attention to the road," they're fake.

 

[diplomat33]

It seems like a pretty straightforward defence to me, being blown out of proportion.

They are suing claiming Elon's statements led the deceased to believe he could play games on his phone and not pay attention to the road. Elon made no such statements about the abilities of Autopilot at the time of the crash. Therefore if there are any recorded statements of Elon saying "You don't need to pay attention to the road," they're fake.

Except that the statements in question come from a 2016 code conference with 500 attendees. They are not fake.

 

[willow_hiller]

Except that the statements in question come from a 2016 code conference with 500 attendees. They are not fake.

The point of the defense isn't what Elon said, it's the fact that Elon is no longer in complete control of what others think he said. Their argument is that there's a sufficient amount of misinformation that it's not possible to know which exact statements influenced Walter Huang to misuse Autopilot, and whether those statements were actually spoken by Elon.

The statements quoted by the prosecution are real, you can listen to them in this recording of the conference:

But how does the prosecution prove that Walter Huang was convinced to misuse Autopilot because of those particular statements, and not other statements that might have been taken out of context, altered, or faked?

 

[Genie]

“Some” EM statements don’t age well

Elon Musk Today

elonmusk.today

 

[KArnold]

Elon Musk announced he was building a new litigation department at Tesla for the purpose of suing people after he was accused of offering a horse in exchange for sexual services. (5/20/2022)

I wonder what the horse said?

 

[tdiggity]

Tesla lawyers claim Elon Musk’s past statements about self-driving safety could just be deepfakes

A new way to avoid liability: the deepfake defense

www.theverge.com

 

[tdiggity]

Surely all of his tweets and public statements in person at his events, arent deepfaked.

/s Who knows who typed the tweet! Could have been an assistant gone awol. Or AI controlled.

 

[Goose66]

The point of the defense isn't what Elon said, it's the fact that Elon is no longer in complete control of what others think he said. Their argument is that there's a sufficient amount of misinformation that it's not possible to know which exact statements influenced Walter Huang to misuse Autopilot, and whether those statements were actually spoken by Elon.

The statements quoted by the prosecution are real, you can listen to them in this recording of the conference:

But how does the prosecution prove that Walter Huang was convinced to misuse Autopilot because of those particular statements, and not other statements that might have been taken out of context, altered, or faked?

That’s completely missing their point. This is not a criminal trial; there is no prosecution; they aren’t trying to present “reasonable doubt,” they are not hanging their hat on reliance on any particular one of Musk’s statements; and they don’t have to “prove” anything but simply convince a jury that more likely than not Huang was using the car as a reasonable person would, the car’s systems failed to operate in a safe manner, and that the company knew such a failure could occur. I think the fact that Autopilot was labeled “beta” pretty much covers the latter two, so the former is probably what will be at issue.

Here the defense was arguing that Elon should not be subpoenaed to give a deposition because any statements upon which he may be questioned about could have possibly not actually be made by him. This argument was presented to a judge in a preliminary hearing, and I think the subsequent articles with judge’s statements certainly reflect the fact that the defense is full of *sugar*.

 

[EVNow]

Definitely. The "deepfakes" defense seems like a desperate move to do something because they have no real defense. Elon's statements are public domain and clear as day. They can't refute them.

Nothing to do with defense... They were trying to prevent Elon being forced to testify. All companies try to keep their execs out of court room. Of course the argument is novel but we'll hear more and more about it in coming days.

 

[Mark II]

And what about the live presentations like Autonomy Day, AI Day and shareholder meetings? Elon was in-person on stage when he made comments about FSD. Those cannot be deepfakes. This Tesla legal defense is truly absurd and bizarre.

Were you actually there? Did you see him in-person? I mean the entire presentation may have been a deepfake live stream! Just saying