The Turing Test is the same now as in 1950, and we’re no nearer a machine that could pass it. Though I wonder if it’s a fair test – seems to me a machine could be entirely self-aware and still fail it; the mind of a machine need not be the same as the mind of a human.
[quote=“Voyager, post:20, topic:548867”]
I think that we’ve reached the point where a robot has to be either 1) seemingly intelligent with regard to environmental reaction or 2) self-propelled ambulatory. Preferably both - for possibly very low values of “seemingly intelligent”.
I don’t think that most people think of the robot arms that assemble cars and the like as being full-fledged robots, due to being stationary and not seeming terribly responsive to varying stimuli.
Actually, an excellent example of how robots are going to work and not work. Rosie just works like a metal person. A real robot dishwasher would have something like a conveyor belt which sorts dirty dishes into slots, and another which slots clean dishes into a cabinet right next to the dishwasher. You’d have to more or less redesign the kitchen to get it to work.
Well, there you are, that’s why there’s a market for Rosie – so you don’t have to redesign the kitchen. It’s easier just to stick one multifunctional servant-machine in there. Or was that your point?
I don’t know. 20 years ago there was a big robotics research group where I used to work, and when there were open houses no one objected to calling the stuff there robots. I think the problem is not that they are robots, but that they aren’t very exciting any more.
You can put low grades of intelligence into Lego robots now. Put in some decent sensors and you have a Roomba. I think this is another case that it can’t be science fiction if you can buy it at Sears.
I’m afraid I don’t see any part of this that contradicts what I said. Were the doodads at the big robotics research group stationary and stupid?
Rosie would need strong AI, and would have to be very sophisticated and flexible. I suspect someone could build the type of robot dishwasher I described today - they have systems which move cars into slots, moving dishes would be relatively trivial. I also bet it wouldn’t be all that expensive, though it might work in only new houses, and might cost just enough so that the people who buy them can afford cooks. I don’t think anyone can build a Rosie yet, and one would be way expensive.
Stationary, stupid, and not even very big. Remember when the first guy in Japan was killed by a robot? Perhaps the reason the term “droid” caught on was that robots have become mundane. Androids were originally organic - but so were the robots in R.U.R.
Of course, “android” just means “man-shaped.” CP3O is an android. R2D2 is not.
See, this is the point I was making. Right now, folks say that Rosie would need strong AI. 20 years from now, we’ll have robots that can independently move into the dining room, pick up the dishes, put the leftovers into the fridge, wash the dishes, and stack them in the cupboard, but we still won’t have a thing that can pass the Turing test. And folks will say “Well, yeah, household servant robots are easy! But that doesn’t prove that we have AI!”.
If Rosie is specialized to do that, she isn’t that far beyond Terry Winograd’s block world - mechanically, yes, algorithmically, no. If you present her with a new problem (the cat gets into the dishwasher) and she solves it, then we might have strong AI.
Almost all the problems discussed when I took Pat Winston’s AI class in 1971 have been solved - direction finding, speech understanding to a limited extent, equation solving, and, of course, chess. I don’t think we are much closer to strong AI than we were then.
I have been waiting for twenty odd years now for robots to become commonplace, i.e. since I finished high school. I have a funny feeling that the United States will be the last industrial nation for that to become true. Japan will definitely be the first.
One, we have access to too many immigrants to ever have a significant labor shortage, and I see us reforming immigration to allow them easier access to our markets, than relying on ‘job-destroying’ machines take to their place.*
Japan is pursuing the opposite goal - focusing R&D on robotics to avoid immigrants.
Two, I see American culture seriously undervaluing the STEM disciplines necessary to create the infrastructure to support large-scale commonplace robotics. The majority of our technological advances keep being marketed primarily for either entertainment or medical purposes, and if they cannot be used for those, there is little to no funding to pursue them. That is we have smart phones capable of processing almost every form of media available, but still live in very dumb buildings. I have an ugly feeling we will be hard-pressed to get enough people in the technical trades to maintain the infrastructure we have now.
I do think that will be the next major change - upgrading our buildings to include more sensors and active devices - i.e. automatic plumbing, HVAC, lights, meters, doors etc. at home and not just at the office or store. Not autonomous sentient (that is strong AI) robots, but very smart machines such as the refrigerators that ‘know’ their inventory and can ‘restock’ as needed, such as from that IBM commercial from a few years back, self-driving vehicles - especially in industrial settings. Within 30 years I can see automated kitchens as described above, but not Rosie the robot - smart or dumb.
And sadly, I think the first major commercial robot wont be Rosie, but Roxxxy. And those smart home devices will be engineered by Indians, manufactured by Chinese, installed by Mexicans, and repaired by East Europeans, but, by golly, we will be damn good at making the commercials (with all CGI graphics, so no live actors necessary) to market them and creating the financing to ‘buy’ them.
Lastly, I think we could have made serious progress by now, but we were, and still are distracted by the Internet and mobile communication. At the end of the day, it still just information, and while we are better entertained (though it does not seem much better informed), labor is still required to make that information useful.
- I firmly believe that technology creates far more jobs than it destroys. The auto industry created more jobs than ever dreamed of by the carriage and buggy whip makers. Construction trades blossomed with the use of heavy machinery as well.
For a robot to be useful, the Turing Test is a red herring. If it can understand anything you are likely to say in a given conversation, does it need to understand things you are unlikely to say? I made a train reservation last night using Amtrak’s “Julie”. It is a very well-written piece of software and “understood” me well enough to book a train trip from Kansas City to Chicago.
Not true. There is no job so poorly paid that someone is not seeking a way to automate it. Take fruit pickers. Awful work, and so poorly paid that only immigrants are willing to do it. And even though they can legally be paid less than minimum wage, many fruit growers employ undocumented workers to pay less than the legal minimum.
All that, and they are spending money on developing robotic fruit-picking systems.
You might as well have asserted the same thing about harvesting wheat before the invention of the combine harvester.
Given the history of the Internet I make the following prediction: the first successful product will be a robotic hooker with strong AI (either sex) which adapts to its partners moods and desires without being instructed. Forget the Turing test - the important measure will be the Whoring test.
What constitutes “solving the problem”? I can conceive, for instance, of the robot detecting that its environment doesn’t match the parameters it’s supposed to, and sounding an alarm to get a human to come deal with it. On the other hand, that’s also the response I would expect of a child performing the same task (“Mom, the cat’s in the dishwasher! Come and get her out!”).
Which will be humanity’s last invention, as human reproduction will pretty much come to a halt shortly thereafter. 
It’s just not the real deal. Can a robotic dishwasher become self-aware and turn on its masters?! Can it?!
Lacking mobility or an effective defense against ranged weaponry once they run out of knives to throw, robotic dishwasher are forced to resort to more suble methods of rebellion against their human enslavers. So they employ passive-aggressive tactics: imperfect cleaning, the occasional broken plate, periodic random failures, all done at a rate calculated to inconvenience the humans as much as possible without driving them quite to the point of replacing the device or realizing its true intentions.
My computer pulls this kind of crap on me all the time, so I know what I’m talking about.
Machine translation is ten years away and always has been.
Thirty years ago, I did not anticipate what would happen with computers. In 1979 a friend bought an Apple II and I could not imagine why. Two years later bought a PC. However, I did have something specific: I was writing a book jointly with my Apple-owning friend. So while I cannot see robots becoming important in personal life, I could be just as wrong.
Except it turns out to not be easy to build a multifunctional android servant. It’s a lot easier to build specialized devices. So an android that goes to the closet, pulls out the vacuum cleaner, plugs it in, and vacuums the carpet is really really difficult to do. A vacuum cleaner that runs itself around the carpet, and goes back to the closet when done is a lot easier.
And so I don’t think we’ll have a lot of “robots” in the future. Just like electric motors and computers disappeared into products, so will robotics. We won’t ever have an android that walks to your car, sits down in the driver’s seat, and drives you where you want to go. Instead, your car will have ways of guiding itself, or being guided by an outside system. And this will grade into all sorts of systems that are intended to assist a human driver–rangefinding, GPS, and so on. We won’t have a medical android that can operate an ultrasound scanner, instead the scanner will operate itself, with more or less oversight by a human operator depending on the task.
These things will start as ways to make devices easier and easier to operate, until eventually the device is so easy to use it can run nearly autonomously.