Condottiere
Emperor Mongoose
Railways are easy, you can't really deviate from the route; the drivers are included to allay passenger anxiety, and if something unexpected happens that requires human intervention.
Artificial Intelligence
- Thread starter dragoner
- Start date
Tom Kalbfus
Mongoose
I think most people would prefer their robotic taxi without the smartalec talking plastic manaquin in the front seat. This is the one from Total Recall, isn't it.Condottiere said:Railways are easy, you can't really deviate from the route; the drivers are included to allay passenger anxiety, and if something unexpected happens that requires human intervention.
Condottiere
Emperor Mongoose
Yes, there were different screenshots, but this one seemed best demonstrative of the concept.
An inner city automated taxi wouldn't go fast, and wouldn't be robustly built.
An inner city automated taxi wouldn't go fast, and wouldn't be robustly built.
Tom Kalbfus
Mongoose
Actually an automated taxi would be a flying drone that carries passengers in all four seats.
Condottiere
Emperor Mongoose
I think that customer's will inform a central computer their requirements, and most will end up with a smartCar.
A
Anonymous
Guest
If all we're using AIs for in YTU is taxis, I'm most opposed to playing in it!
:mrgreen:
Really, with the potential of an AI there are many ways they can affect a society that are deeply more significant than taxis.
:roll:
Did the thread die?
Of course it did, Tom's the thread killer
:mrgreen:
Really, with the potential of an AI there are many ways they can affect a society that are deeply more significant than taxis.
:roll:
Did the thread die?
Of course it did, Tom's the thread killer
Condottiere
Emperor Mongoose
I think we're discussing the pragmatic applications.
A
Anonymous
Guest
In which case, I can see plenty of AI suicides... "Brain the size of a planet and all I get to do is run the 57th century Uber"
It doesn't take AI to run an automated transit system, as has been pointed out, we have the ability to do it today, we're just waiting for the lawyers to argue amongst themselves for it to become a legal/commercial venture.
We are not close to building an AI in the context of the OP.
It doesn't take AI to run an automated transit system, as has been pointed out, we have the ability to do it today, we're just waiting for the lawyers to argue amongst themselves for it to become a legal/commercial venture.
We are not close to building an AI in the context of the OP.
Tom Kalbfus
Mongoose
Honestly, would you rather have a smartCar or a flying car? smartCars look ridiculous, they are basically a couch on four wheels, and it don't looks safe, what happens if you get in an accident with a truck? Your a little better off than if you were on a motorcycle, but not much better than that. Flying cars would work if you had a computer controlling the movements of all such vehicles in the sky so as to avoid collisions. My guess is all flying cars would be networked drones controlled by a cab company or the FAA, maintenance requirements would be rigourus and the cars would be expensive because of those requirements. So you hire them to take you from point A to point B much as you would use an Uber cab.Condottiere said:I think that customer's will inform a central computer their requirements, and most will end up with a smartCar.
Reynard
Emperor Mongoose
I'm seeing a trend on this thread, people are misusing the descriptor of A.I. for what are actually, at the extreme, expert systems. Many are closer to sophisticated dumbots including possibly Jonnycab though the eye rolling.... You want a scifi A.I., try K.I.T.T., Jonny Five, Gadgetmobile or any silly cartoon show that needed a computer with an awesome personality. Cute, fun and what people see as A.I.. Some might even remember HAL.
Good example for separating robots with expert system and A.I.s is Will Smith's version of I, Robot. The USR robots are all robots with expert systems. VIKI and Sonny are actual A.I. representing the building sized with huge command and control capacity and a brain in a mobile shell, respectively.
Good example for separating robots with expert system and A.I.s is Will Smith's version of I, Robot. The USR robots are all robots with expert systems. VIKI and Sonny are actual A.I. representing the building sized with huge command and control capacity and a brain in a mobile shell, respectively.
Condottiere
Emperor Mongoose
You could add in a DNA programming strand that artificially constrains performance beyond a certain point, and/or firewalls the AI from certain connections.
Probably can be overcome with time, but allows an easy digestion of information without overwhelming it, and preventing it from taking certain actions that might be harmful to itself or us.
Probably can be overcome with time, but allows an easy digestion of information without overwhelming it, and preventing it from taking certain actions that might be harmful to itself or us.
CosmicGamer
Mongoose
No, the trend is to misuse the descriptor of AI as the entertainment/sci fi version where the AI is not "artificial" "intelligence" and instead something with, as you say, personality.Reynard said:
As I keep saying, curiosity, imagination, personality, will to live, and so on are human traits other than intelligence.
Condottiere
Emperor Mongoose
The AI might be city manager, or just run the transport network.
hiro said:
How about taxis that are everywhere, are super cheap and are so efficient and safe that private car ownership ceases to exist outside of hobbyists. Even then human drivers might only be allowed on private roads.
Reynard said:
Precisely, and in fact that's what Stuart Russel is talking about. AI research these days is all about special-purpose expert systems that are good at producing optimal solutions to specific problems but which have no general purpose intelligence.
Simon Hibbs
CosmicGamer said:
I'd say there's a trend towards both. Humans have a very powerful tendency to anthopomorphize things.
Where I will quibble with you though is curiosity, and perhaps to some degree imagination. Even extremely smart single-purpose optimizing expert system type AIs have no need of these, but any general purpose strong AI is going to need to be able to pursue apparently incidental or superflous lines of reasoning or inquiry, because sometimes they will turn out to have unexpected value. I think it would be resonable to characterize that as curiosity. Similarly, if imagination is the ability to hypothesize possibilities and generate and explore variant conceptual models is a valuable or even necessary approach to general problem solving then we might characterize that as imagination. Even some classes of advanced specialized problem solvers might need at least vestigial forms of these capabilities.
Let's take some of the examples of AI that Stuart Russel is very impressed by in the article Reynard posted. The image classifiers are a good one. These things are extremely efficient at classifying breeds of dog for example. But these systems have absolutely no conception of what a dog is, how they behave, or even what they are like as 3D objects. All they know is how to classify two dimensional matrices of arbitrary binary data - pictures. In fact I've seen examples of classifiers being fooled by very small adjustments to the images. For example a classifier might have a 99% chance of identifying cars, apples and banaas, but very subtle changes to the binary data in the photo of a banana that clearly leaves it unambiguously a banana to human classifiers can lead the computer classifier to think it's a car. The classifier has no idea what a car is, or a banana. A human observer can imagine the banana as a 3D object, can see the context of the banana as lying on a plante on a kitchen counetr top. The image classifier doesn't know anything about plates or counter tops. It's operating in a completely contextless fashion.
These things are only intelligent in the sense that they can solve problems, but they are incredibly specialised to the point of total uselessness for anything other than the very specific problem they're designed to solve.
I think general purpose strong AIs absolutely will need some attributes in common with human intelligence in order to be effective, but not necessarily others. Imagination and curiosity are two good candidates, but I completely agree with you about personality and will to live. I'd also add things like any emotions such as anger, desire, love, even emotional states like loneliness. Those are entirely superfluous to whether a system is a general purpose intelligence.
Simon Hibbs
Similar threads
