Tesla Autopilot 'a distant second' to GM's Super Cruise in hands-free test | Hacker News Reader

Tesla Autopilot 'a distant second' to GM's Super Cruise in hands-free test

reuters.com

27 pointselorant11 comments

I bought a subscription to CR when I was going to buy a TV. They pick random categories to use when deciding which one is "best". This TV has a large bezel but is perfect in every other way? Well too bad 80/100 and the super expensive Samsung gets a 92/100.

So there is probably something better about GM, but it in no way means you will be happier with it compared to FSD

0 deaths would rank pretty high on my "reasons to be happier with it".

There are lots of claims that Tesla self-driving features are dangerous, is there any data to support this?

The fact that their detection of driver attention is worse is only one factor, the cars should be evaluated on overall accident rates relative to human drivers or other self-driving systems.

This is obviously wrong. https://twitter.com/WholeMarsBlog/status/1321732764303990784

I think that this article is nothing more than an attempt to manipulate the stock prices in question.

Consumer Reports and Reuters don't attempt to manipulate stock prices. You might disagree with their conclusions but you're obviously wrong about their motivations.

You might very well be right, in that those specific organizations likely have no hand in or knowledge of manipulating stock prices. (I wish I could go back to the original comment and tweak the wording to reflect this.)

But what do you think about others spreading the articles selectively to get more reach and thus influence what common shareholders do? That's a pretty easy way to achieve the exact same effect, and a lot harder to trace.

You can learn a lot by observing the gaps...

CR has had some badly designed tests in the past, but the way they run their org is outstanding. They're a nonprofit, they purchase all products retail, they accept no advertising dollars, they forbid using their testing results in advertising.

It seems natural and very above board for Cruise to want to spread this good review, though they cannot use it in advertising.

> DETROIT (Reuters) - ...

AP is a "distant second" because the Consumer Reports scoring mechanism weighs the criteria of "babysitting" the driver's attention as much as the actual autonomous driving performance. If you look at the scores, GM is a better nanny, while Tesla is better at the actual autonomous driving task.

> In recent European safety testing, a Tesla Model 3 with Autopilot placed sixth out of 10 systems, getting high marks for performance and ability to respond to emergencies, but falling short on its ability to maintain a driver’s focus on the road.

This undermines my confidence in the entire rating system. At the very least, the drivers focus should be its own rating distinct from the performance of the piloting.

The serious selling point of any of these systems is greater-than-human reactions in emergency situations, and so a buyer going off these ratings could be mistakenly lead to believe that they are getting a better system in a Cadillac when they flat out aren’t.

Objectively, SuperCruise is better than Autopilot. CR noted that the performance of SuperCruise on the self-driving tasks was almost as good as it was in Autopilot.

The big difference was that the safety functions on Tesla were subpart. As CR rates products based on their overall scores, this resulted in SuperCruise taking the crown.

TLDR: Ratings aren't just about one feature. Everything matters.

Reply on news.ycombinator.com