Kronopath

I want to push back against your pushback as someone who’s lived in both NYC and the SF area. I agree with you that Uber barely made sense in Manhattan. I never once used it and taxis were plentiful.

I’ve since realized that in the US, NYC is an exception. When I first came to SF and Seattle for job interview related things, I used taxis, only to find out that the taxis were so terrible I never used them again:

- In the suburbs of Seattle, I was given a taxi chit from the place I was interviewing. I called in for a cab and had to wait over a half an hour for one to pick me up.

- In SF, the airport cab I was using had his GPS unmounted from his dash, and ended up handing me the machine and asking me to help him navigate from the back seat. Then when we got to the hotel, he lamented my choice to pay by credit card as it meant he would get the money much later than if he had cash. After I told him I didn’t have the circa $100 in cash he was charging, he sadly acquiesced, then proceeded to take a literal paper rubbing imprint of the card number before I could leave.

I like to say that the Bay Area made Uber make sense, both in terms of urban planning and in terms of how terrible taxis were.

And I think those may be related: if you can get around well in a place like NYC using public transit or walking, taxis have to be a lot more attractive in order to justify their existence. In SF or Seattle they had much less competition due to the suburban sprawl and worse public transit.

Anything that allows AI to scale to superinteligence quicker is going to run into AI alignment issues, since we don’t really know a foolproof way of controlling AI. With the AI of today, this isn’t too bad (the worst you get is stuff like AI confidently making up fake facts), but with a superintelligence this could be disastrous.

It’s very irresponsible for this article to advocate and provide a pathway to immediate superintelligence (regardless of whether or not it actually works) without even discussing the question of how you figure out what you’re searching for, and how you’ll prevent that superintelligence from being evil.

Yes, but it’s not always perfect.

I’ve been notified of an “unknown AirTag” while I was home. When I checked the locations it was seen with me, it was a random zigzag within a block or two of my home.

I’m pretty sure what happened is that the AirTag belonged to one of my neighbours, there were some GPS distortions happening that made my phone think it was moving slightly, it kept hearing the AirTag’s signal, and it assumed I was being stalked while wandering near home. This person might have the same thing happening to them.

A better comparison for numerical classifiers would be uncountable nouns.

In English, you don’t say “give me three waters”, you say “give me three glasses of water” or “three bottles of water”. You can think of the classifier words as being that, but for everything:

三杯水 three glasses of water

十头牛 ten heads of cattle

两支铅笔 two rods of pencil

一条路 a strip of road

六只猫 six animal-units of cat

五个人 five “gè” (generic units) of people

No one should want this.

This is not good news, this means that we could end up with a dangerously superintelligent AI just by scaling up the number of parameters, without increasing the amount of training data.

Legally, it’s not compulsory. Socially and culturally, it definitely is. In the US, if you don’t tip 15%, you’re either an asshole, or you’re saying there’s severe issues with the service.

I’m not going justify this culture—I don’t like it either—but that is the way it is.

Yes.

In SF, up-front prices have been comparable to Uber’s, except that you don’t have to pay tip, which automatically makes it 15–20% cheaper.

The few times I’ve tried it, the service has been good and its driving was safe. The only downside is that there seems to sometimes be longer wait times.

Does this mean they can go on the highways? If so that’s a significant upgrade to the Waymo service in the SF Bay Area, as it can now take you between towns.