73ms

I have significant doubts about this “any Tesla as robotaxi” idea at least if you’re referring to Musk’s empty promises about how you could make money off your own Tesla while not using it. The simple explanation is indeed that they’ve noticed 2 seats is enough for most taxi usage and IIRC I’ve seen some articles say the data supports that. They’ve optimized the Cybercab to be their taxi in this and other ways. Clearly they’ve also made efforts to maximize its range and chosen an interior that is easy to clean, removed the glass roof and even the rear window that would possibly not last so long in heavy use etc. If there’s ever going to be a bigger taxi from them I’d expect it to be another purpose-built vehicle.

doesn't really make that much sense to look at the registered user total because most of them have already moved on but sure, it's like 2.5% if you compare to the 5 million that have been active within the last month.

I think the ecosystem has some more benefits besides that. Ready to use case, cooler, camera, display options and lots of resources for putting these things together compared to NUCs/minipcs. I do like how the compute modules are even more flexible with the ability to choose from various carrier boards. Really depends on your use case which makes the most sense.

Did this say whether the reasoning models get this right more than the others? Was curious about that but missed it if it was mentioned.

Yeah I do think there’s real utility to these and they’ll become more and more useful. These companies are working towards AR glasses that could completely replace phones.

Well, according to them we already have them: waymo.com/safety/impact/ There are some limited third-party studies also supporting this but to be fair I think there’s still more needed for independent verification. Of course, it also matters what exactly we consider safer, it is entirely possible that they could be the cause of more minor damage while also preventing deaths that would occur with humans at the controls.