Did this lady not have hands? Because I could see how a lady without hands or arms using Siri to call for help would be newsworthy. Perhaps there was some other reason why she was incapacitated and could not simply dial '911'? Otherwise, what service has Siri provided here, other than saving a second and a half to unlock the phone and dial?
I just commented the same above. I haven't tried all the radio services.. for one thing I got turned off of Spotify because of the Facebook relationship. None of the services seemed to match by song quality.. Pandora was the only service that gave me other songs that were good accompaniments with each other. The other sites just seemed to go by artist genre.
I've made playlists in other radio sites and they just seem to play other artists in the same genre which is never really what I want. Pandora was the only one I knew which would actually match by the qualities of the songs I chose regardless of the artist, therefore it was the only one I used.
Statistically speaking you are correct. The problem is people get lazy. Everyone is capable of driving safely but then life happens and they forget to shoulder check and they get in an accident. Personally speaking, my driving record is better than AI because I never forget to shoulder check. Furthermore, there are many people out there who are safe drivers and will benefit little from AI. In an accident where two or more people were injured, it is one person doing something unpredictable and I see no evidence of current AI being able to handle unpredictable situations. So while we may be able to help driver safety by making AI mandatory for bad drivers, mass adoption would be overkill in that regard because many people already drive safely.
Can we agree that AI should be able a computer's ability to deal with situations where there are not well-defined rules? I don't even include Go as an example of AI because the rules for Go are easy to understand. Driving may be a better example, since there are edge cases not covered by a driver's handbook. However, I have yet to see any real word example's of Google's AI being able to deal with situations that go beyond a driver's handbook.
Go is not the hardest thing to do. There is a set of well-defined rules for go. In fact, the rules for Go are simpler then for Chess or Poker. All that has changed is our ability to capture the massive complexity of those rules on a Go sized board with raw computing power. If we know all the rules then there are no edge cases, and AI is all about the edge cases that occur in the natural world. AI is precisely about a lack of rules and a computer's ability to deal with them.
Agreed. Once you work a lot on all these languages, Python feels a lot more like pseudo-code and therefore more natural. Java has some things you need to do to satisfy the grammar of the language itself, and more so for Objective-C. This is good in certain larger projects with many people, but it depends what you want to do.
It's funny you say that, because I think Google's driving AI just has short term memory of the small area they're driving on which, yes, just amounts to lookups. I think if they pick them up and move them to another city they would be almost back to square one.
That goes on your insurance, but that is nothing do do with a driving record, therefore that component of insurance is much cheaper. If insurance companies are willing to cover to property damage to my car then fine, but I don't want it connected to my driving record like it is today and I don't want higher premiums in the case of an accident. When I consider the value of my house and the amount that my property insurance costs, then insurance for the property of my car should cost around $100 a year. If that's what AI insurance costs then fine, but that is nowhere close to what my current vehicle insurance costs.
I develop in python, java, and objective-c... you know what I hate more then python? Putting parenthesis around every single conditional statement. Once you work in python for awhile and go back, you start to notice how many brackets there are in the other languages.
The person who puts it in a car and hooks it up to the steering wheel as a complete 'automated driving system' should undergo rigorous testing in phases, similar to drug trials. Then they should be responsible for all damages caused by the AI.
That's the thing.. insurance agencies rely on the fact that you will not want penalties and will therefore drive slower and more carefully. Maybe you get in an accident, then you drive even more carefully; finally you lose your license. An AI car just is what is is. And if there is no one in the car, who is liable? The owner, who has nothing to do with the behavior or performance of the car? The insurance industry doesn't want to fight Google.
Oh and about 'it can drive a car'. Perhaps they have a core set of learning that can drive a car under certain conditions in the real world. But they are still testing under very specific and controllable conditions. Isn't Google testing in California? One of the least changing climates in the world. All the edge cases are elsewhere, and it's the edge cases that are killer for AI. When they can drive an AI car on a road under 12 inches of snow and in a blizzard, then it will be ready for mass market.
Meh, they were talking about neural nets back when I was in school. If it has only progressed from Chess to Go in that time, then it has a long way to go.
There is no such thing as 'weak' AI. It is either AI (something seeming human and thus can do many human things) or it is not. The AI that can win at go is so laughably dedicated to playing the game of go it is nothing more than a gimmick.
Slashdot Mirror
Why was it out of reach? Was she tied up? Is she missing arms and legs? Did the phone fall into a gaping chasm??
Did this lady not have hands? Because I could see how a lady without hands or arms using Siri to call for help would be newsworthy. Perhaps there was some other reason why she was incapacitated and could not simply dial '911'? Otherwise, what service has Siri provided here, other than saving a second and a half to unlock the phone and dial?
I can see it now. "Oops, your scheduled ride came up during a period of high surge pricing! That will be $100 please."
We should automate them immediately. People say AI is getting pretty good. That's enough.
I just commented the same above. I haven't tried all the radio services.. for one thing I got turned off of Spotify because of the Facebook relationship. None of the services seemed to match by song quality.. Pandora was the only service that gave me other songs that were good accompaniments with each other. The other sites just seemed to go by artist genre.
I've made playlists in other radio sites and they just seem to play other artists in the same genre which is never really what I want. Pandora was the only one I knew which would actually match by the qualities of the songs I chose regardless of the artist, therefore it was the only one I used.
You're right, it's only -25C pre-windchill.
Where I am it goes down to -35C.
I am told that EVs handle this temperature just fine, because they use EVs in Norway.
Yeah I can't really get that excited about a programming language either way. I just use the tool for the job.
yes, and I meant to say brackets. Am I not allowed to increase the scope of my comment in the next sentence?
Statistically speaking you are correct. The problem is people get lazy. Everyone is capable of driving safely but then life happens and they forget to shoulder check and they get in an accident. Personally speaking, my driving record is better than AI because I never forget to shoulder check. Furthermore, there are many people out there who are safe drivers and will benefit little from AI. In an accident where two or more people were injured, it is one person doing something unpredictable and I see no evidence of current AI being able to handle unpredictable situations. So while we may be able to help driver safety by making AI mandatory for bad drivers, mass adoption would be overkill in that regard because many people already drive safely.
Can we agree that AI should be able a computer's ability to deal with situations where there are not well-defined rules? I don't even include Go as an example of AI because the rules for Go are easy to understand. Driving may be a better example, since there are edge cases not covered by a driver's handbook. However, I have yet to see any real word example's of Google's AI being able to deal with situations that go beyond a driver's handbook.
Go is not the hardest thing to do. There is a set of well-defined rules for go. In fact, the rules for Go are simpler then for Chess or Poker. All that has changed is our ability to capture the massive complexity of those rules on a Go sized board with raw computing power. If we know all the rules then there are no edge cases, and AI is all about the edge cases that occur in the natural world. AI is precisely about a lack of rules and a computer's ability to deal with them.
Statistical models and pattern matching.
Agreed. Once you work a lot on all these languages, Python feels a lot more like pseudo-code and therefore more natural. Java has some things you need to do to satisfy the grammar of the language itself, and more so for Objective-C. This is good in certain larger projects with many people, but it depends what you want to do.
It's funny you say that, because I think Google's driving AI just has short term memory of the small area they're driving on which, yes, just amounts to lookups. I think if they pick them up and move them to another city they would be almost back to square one.
That goes on your insurance, but that is nothing do do with a driving record, therefore that component of insurance is much cheaper. If insurance companies are willing to cover to property damage to my car then fine, but I don't want it connected to my driving record like it is today and I don't want higher premiums in the case of an accident. When I consider the value of my house and the amount that my property insurance costs, then insurance for the property of my car should cost around $100 a year. If that's what AI insurance costs then fine, but that is nowhere close to what my current vehicle insurance costs.
What scares me is that Google keeps saying their close, but from what I see in the news they are nowhere close.
I develop in python, java, and objective-c... you know what I hate more then python? Putting parenthesis around every single conditional statement. Once you work in python for awhile and go back, you start to notice how many brackets there are in the other languages.
The person who puts it in a car and hooks it up to the steering wheel as a complete 'automated driving system' should undergo rigorous testing in phases, similar to drug trials. Then they should be responsible for all damages caused by the AI.
.. want the cure to cancer.
That's the thing.. insurance agencies rely on the fact that you will not want penalties and will therefore drive slower and more carefully. Maybe you get in an accident, then you drive even more carefully; finally you lose your license. An AI car just is what is is. And if there is no one in the car, who is liable? The owner, who has nothing to do with the behavior or performance of the car? The insurance industry doesn't want to fight Google.
Oh and about 'it can drive a car'. Perhaps they have a core set of learning that can drive a car under certain conditions in the real world. But they are still testing under very specific and controllable conditions. Isn't Google testing in California? One of the least changing climates in the world. All the edge cases are elsewhere, and it's the edge cases that are killer for AI. When they can drive an AI car on a road under 12 inches of snow and in a blizzard, then it will be ready for mass market.
Meh, they were talking about neural nets back when I was in school. If it has only progressed from Chess to Go in that time, then it has a long way to go.
There is no such thing as 'weak' AI. It is either AI (something seeming human and thus can do many human things) or it is not. The AI that can win at go is so laughably dedicated to playing the game of go it is nothing more than a gimmick.