@kuoh I often have to remind myself that the goal is not to answer it correctly or accurately, but rather to answer it the way a majority of humans answer it.
And it’s not even a “half of everybody is dumber than average” thing — it’s simply that I assume most people are less careful than I am. So I usually pretend that I don’t see the little slivers bleeding into the next box. It seems to work most of the time, even though it hurts me inside.
@kuoh@phendrick@xobzoo when you said FSD, I didn’t know if you meant Full Self Driving. I have tried it and it’s really scary but always seemed to do safe stuff. Will wait for a few more updates.
To further a discussion later in this thread about “smarter than people,” I didn’t know I needed to shop for a Turing car. (Sorry, couldn’t resist!)
@phendrick@pmarin@xobzoo Elon had originally “projected” real FSD to be ready around 2019-2020, not the almost there but not really version they have today. It has come a long ways and even cost a few lives in the process, but definitely still has many more iterations to go before we can comfortably take a nap on the commute to work.
Considering they’ve only just now perfected accurate AI recognition of a single well standardized traffic control signal, imagine how much longer it will take before they can develop an AI competent enough to handle daily driving in all weather conditions, without artificially limiting vehicle speeds. Even then you might still have to worry about its ability to handle the Trolly Problem without going full HAL 9000 on you.
@kuoh@phendrick@xobzoo after the latest SW updates which can come over the air whether you want them or not, it is very bitchy even if not in FSD (not paying at the moment) but use Autosteer which is similar on a major road (without city navigation) I can’t do anything more than about 2-3 seconds before I get a “Please pay attention to the road ahead!” And it gets more scolding very quickly. I don’t even know all the angles it’s watching me from. Plus need to jiggle the wheel every 45 seconds or so. The old stories about drunk people passed out while car drives them home are not happening anymore (that’s probably a good thing overall).
I think it did save my a$% last week in one of those sudden congested traffic situations where everyone goes from 50mph to dead stop and I was looking away for I think just a few seconds which is all it takes. Honestly I don’t think it cared about my a$% but it cared about its front bumper, and apparently thought don’t care what you think, I’m stopping NOW! And I’m very grateful.
I was one of those people who, very early on, predicted that self driving cars work gonna be here much more quickly than people thought.
Within a year, I fully altered that prediction to be the polar opposite: That we were very far away from actual self driving cars.
To wit, we have only barely gotten to some of the major technical achievements. We may get much of the rest and fairly short order… who knows? But we really even haven’t started any dialogue or substantial work on the ETHICS of self driving cars. is going to be even harder and more contentious than the technical part.
(I am also of the strong opinion that self driving cars are, in essence, merely beta tests for flying cars.)
@kuoh@phendrick@xobzoo with recent SW updates it seems to already have a “map” of all known traffic signals and stop signs. It shows on the screen on your route. So I think it knows to look for them ahead of time now. Also I think it reads road speed limit signs, not just using stored map data. It handled a temporary construction zone with 50 mph speed limit once I passed the sign. But after that it didn’t know that “end construction” meant it could resume standard speed which was 60-65. Silly AI.
Honestly it’s never done anything unsafe and if anything is too cautious, which is probably a good thing. It is more like “I don’t know, dude, you better handle this,” so no napping on your commute or drive home from the bar quite yet.
@haydesigner@kuoh@phendrick@xobzoo well then there is the ETHICS of AI in general. This is not a new issue, academics (whatever those are) have debated this for at least 50 years, it’s just that now the technology and the billions of monkeys typing on keyboards (I.e. all of us adding things to the internet) enabled it (so we are all enablers, I suppose)
The HAL thing was a classic AI ethics quandary. Robotics “law” says do not harm the humans; but what if the mission is secret and prioritized in the SW to be more important ? (Sorry, spoiler; you don’t learn this until 2010)
And @kuoh, sorry for my ignorance, what is the trolley problem?
@haydesigner@phendrick@pmarin@xobzoo An extremely complex and difficult to process ethical thought exercise. The Kobayashi Maru of the pre federation society if you will. This might help.
Select all the bicycles… oh and we won’t tell you we are including a motorcycle you have to select too. The same happens with trucks - a school bus needs clicked. Mountains but click the one with palm trees too. I hope they are not using these to train self driving cars.
@Kidsandliz For some reason, I keep getting a vision of thousands of people at desks in Indian call centers with Tesla camera feeds on a monitor and a Logitech gaming wheel in front of them.
Two basically-related things… (the bear one was mentioned in a comment on the article, but I was already searching through my pictures for it before I even read the article)
As a fellow robot, I understand your pain.
FSD beta 3742.5631 nearly ready.
Elon
@kuoh I often have to remind myself that the goal is not to answer it correctly or accurately, but rather to answer it the way a majority of humans answer it.
And it’s not even a “half of everybody is dumber than average” thing — it’s simply that I assume most people are less careful than I am. So I usually pretend that I don’t see the little slivers bleeding into the next box. It seems to work most of the time, even though it hurts me inside.
@kuoh @xobzoo but to my experience, taking those slivers into account makes it more likely for me to get the craptcha bot approval. YMMV.
@kuoh @phendrick @xobzoo when you said FSD, I didn’t know if you meant Full Self Driving. I have tried it and it’s really scary but always seemed to do safe stuff. Will wait for a few more updates.
To further a discussion later in this thread about “smarter than people,” I didn’t know I needed to shop for a Turing car. (Sorry, couldn’t resist!)
@phendrick @pmarin @xobzoo Elon had originally “projected” real FSD to be ready around 2019-2020, not the almost there but not really version they have today. It has come a long ways and even cost a few lives in the process, but definitely still has many more iterations to go before we can comfortably take a nap on the commute to work.
Considering they’ve only just now perfected accurate AI recognition of a single well standardized traffic control signal, imagine how much longer it will take before they can develop an AI competent enough to handle daily driving in all weather conditions, without artificially limiting vehicle speeds. Even then you might still have to worry about its ability to handle the Trolly Problem without going full HAL 9000 on you.
KuoH
@kuoh @phendrick @xobzoo after the latest SW updates which can come over the air whether you want them or not, it is very bitchy even if not in FSD (not paying at the moment) but use Autosteer which is similar on a major road (without city navigation) I can’t do anything more than about 2-3 seconds before I get a “Please pay attention to the road ahead!” And it gets more scolding very quickly. I don’t even know all the angles it’s watching me from. Plus need to jiggle the wheel every 45 seconds or so. The old stories about drunk people passed out while car drives them home are not happening anymore (that’s probably a good thing overall).
I think it did save my a$% last week in one of those sudden congested traffic situations where everyone goes from 50mph to dead stop and I was looking away for I think just a few seconds which is all it takes. Honestly I don’t think it cared about my a$% but it cared about its front bumper, and apparently thought don’t care what you think, I’m stopping NOW! And I’m very grateful.
@kuoh @phendrick @pmarin @xobzoo
I was one of those people who, very early on, predicted that self driving cars work gonna be here much more quickly than people thought.
Within a year, I fully altered that prediction to be the polar opposite: That we were very far away from actual self driving cars.
To wit, we have only barely gotten to some of the major technical achievements. We may get much of the rest and fairly short order… who knows? But we really even haven’t started any dialogue or substantial work on the ETHICS of self driving cars. is going to be even harder and more contentious than the technical part.
(I am also of the strong opinion that self driving cars are, in essence, merely beta tests for flying cars.)
@kuoh @phendrick @xobzoo with recent SW updates it seems to already have a “map” of all known traffic signals and stop signs. It shows on the screen on your route. So I think it knows to look for them ahead of time now. Also I think it reads road speed limit signs, not just using stored map data. It handled a temporary construction zone with 50 mph speed limit once I passed the sign. But after that it didn’t know that “end construction” meant it could resume standard speed which was 60-65. Silly AI.
Honestly it’s never done anything unsafe and if anything is too cautious, which is probably a good thing. It is more like “I don’t know, dude, you better handle this,” so no napping on your commute or drive home from the bar quite yet.
@haydesigner @kuoh @phendrick @xobzoo well then there is the ETHICS of AI in general. This is not a new issue, academics (whatever those are) have debated this for at least 50 years, it’s just that now the technology and the billions of monkeys typing on keyboards (I.e. all of us adding things to the internet) enabled it (so we are all enablers, I suppose)
The HAL thing was a classic AI ethics quandary. Robotics “law” says do not harm the humans; but what if the mission is secret and prioritized in the SW to be more important ? (Sorry, spoiler; you don’t learn this until 2010)
And @kuoh, sorry for my ignorance, what is the trolley problem?
@haydesigner @phendrick @pmarin @xobzoo An extremely complex and difficult to process ethical thought exercise. The Kobayashi Maru of the pre federation society if you will. This might help.
KuoH
@kuoh @phendrick @pmarin @xobzoo
Nice.
What I hate in those is the inevitable presence of one photo so fuzzy (and often fragmentary) that it’s impossible to figure out what the fleep it is.
https://xkcd.com/1897
@narfcake
@narfcake @werehatrack Look for the tail pipe!
KuoH
Select all the bicycles… oh and we won’t tell you we are including a motorcycle you have to select too. The same happens with trucks - a school bus needs clicked. Mountains but click the one with palm trees too. I hope they are not using these to train self driving cars.
@Kidsandliz For some reason, I keep getting a vision of thousands of people at desks in Indian call centers with Tesla camera feeds on a monitor and a Logitech gaming wheel in front of them.
KuoH
Two basically-related things…
(the bear one was mentioned in a comment on the article, but I was already searching through my pictures for it before I even read the article)