Barumonk

Artificial Intelligence - Part 2; "Rosie the Robot serves pancakes."

Rate this Entry
For those of you born more recently, Rosie the Robot is from an older animated series called the Jetsons. She basically looks like a couple large tin cans with some antenna and claws and acted as the maid for the family. You can actually buy something quite similar from a start up company called Willow Garage.

Willow Garage's PR2 (stands for Personal Robot, Version 2) is a small robot that accepts customer written scripts to do different tasks. Think of it like the Matrix when Neo is getting all that kung fu uploaded into his brain, same idea. This touches on what I said in Part 1 about assembling a bunch of weak AI traits to make a strong AI, this is really it right here. Willow Garage's concept is that you pay less for the robot if you make all your scripts open source. After enough scripts for everything exist, then the robot should be able to do anything and everything with a high level of skill and zero learning curve.

Of course, this isn't a true strong AI because it can't learn from it's own mistakes, but this is really where the industry as a whole is headed right now. There are others doing different things, but they really lack funding. The problem is that most companies want AI for a specific purpose, and a weak AI is faster to implement, easier to control and debug, and costs less than attempting to develop a strong AI - so no one funds strong AI development. What purpose is there to create a strong AI when a weak AI fills the needs of those with money to throw away?

Well, let's think about it. Have you ever played Civilization? If you have, you know that there is quite a lengthy research tree that goes through several eras in human history. So, what if research was done in a different order? What if humans discovered these technologies and made these breakthroughs in a different order? Wouldn't that mean that some technologies would never have been discovered, and that others that we haven't even though of would instead take their place? Yeah, that's right. You could use AI to emulate species and civilization development and potentially create breakthroughs by altering the conditions of the AI's world.

Another concept that extends this, what about alien species? If you tried this with completely different parameters for the Earth itself, then the entire species could evolve and develop quite differently over the course of several million years. It could give us an idea of what to look for instead of going in blind. Playing god has it's benefits, considering the things most companies do, I don't see why they don't seriously consider something like this.

The pancakes are done. Where did I put the syrup?

Comments

  1. Animeniax's Avatar
    Here's a revealing article about the nightmarish world of the Jetsons, and why we should postpone the use of personal robots until we can control them better:

    The Jetsons: They Burned the Sky


    The Jetsons takes place in the futuristic utopia of Orbit City where George, the man of the house, is employed full time at Spacely's Sprockets for a total of nine hours a week. Robots and computers handle nearly all of the grunt work, leaving the bourgeois citizenry plenty of leisure time to shop for such frivolities as multi-dresses and ice cream for their space dogs.


    There's apparently no space-fuel crunch.

    The Jetsons live high above the clouds in their Skypad apartment. In fact, all of the important places in their lives are above the clouds, including George's workplace, the schools and the shopping centers. Wait, why is "in the sky" the safest, most cost-effective place for an elementary school?

    So, What's the Problem?

    The natural environment is gone.

    The surface of the Earth is never shown, and the Jetson family never visits it. They often venture off-world like it ain't no thing, but never down to their own planet. We only have a few stray clues that point to the state of the Earth's surface: In Jetsons: The Movie, Rosie pushes a button to have the Jetsons' apartment rise above the planetary smog.


    We too adhere to the "standing up real high" school of environmentalism.

    And in The Jetsons Meet the Flintstones movie, when George visits the past, he makes an offhand comment that grass is something he "remembers from ancient history."

    When something as ubiquitous and hardy as grass -- something that grows in freezing tundra and burning desert alike -- is "ancient history," the only logical conclusion is that nothing grows on the surface of the planet. It is so polluted, irradiated or burned that no life exists there. The fact that George Jetson hints at the fate of the Earth in a Flintstones crossover actually has even more worrying implications:

    In the Flintstones universe, primitive man enjoys roughly the same quality of life as modern man, but only by virtue of animal exploitation. A camera, for example, is just a box with a bird that pecks the image into a stone tablet, a vacuum cleaner is a woolly mammoth trunk, and so on. The main problem is that these aren't just animals. They're intelligent: They think, speak and joke. They turn to the camera and say things like "It's a living" or some other glib line before dejectedly resuming their "jobs." Jobs that entail extreme suffering and humiliation: The steam whistle at Fred's job, for instance, is a bird. It's activated by yanking its tail until it screams in pain.

    Fred's alarm clock is also a bird: The snooze function is activated by punching its tiny skull in. Just for doing its fucking job and sounding the alarm that Fred himself set.

    If the Flintstones and Jetsons exist in the same universe, just in different eras, and there are no dinosaurs in the Jetsons cartoon, then somehow the dinosaurs from The Flintstones, like ours, have gone extinct. But our dinosaurs were just dumb beasts, and they went extinct long before humans had evolved. In the Flintstones universe, humans and dinosaurs still coexist. They're actually dependent on one another. One is not going extinct without affecting the other. So in the span of time between The Flintstones and The Jetsons, some cataclysmic event occurs that kills off just the creatures, but not the humans. And the Jetsons universe, with its scorched, unusable Earth, hints at what that event might have been: The dinosaurs, like every other creature on Earth not brought into the sky to dance and amuse future man, were either killed off through massive environmental negligence once we finally learned how to replace them with technology, or else, like in The Matrix, we simply burned the sky in a desperate, last-ditch attempt to stop the great dino uprising.


    Their corpses were fuel, and we needed our sport utility oblongs.
  2. Kraco's Avatar
    You could use AI to emulate species and civilization development and potentially create breakthroughs by altering the conditions of the AI's world.

    Another concept that extends this, what about alien species?
    A strong AI would already be an alien species. If it wasn't, it would be just a digital copy of an existing human consciousness and consequently next to worthless for simulating anything novel.
  3. Barumonk's Avatar
    Quote Originally Posted by Kraco
    A strong AI would already be an alien species. If it wasn't, it would be just a digital copy of an existing human consciousness and consequently next to worthless for simulating anything novel.
    Not true, once you complete a strong AI then similar to how you use different patterns in programming to create different results in an efficient manner, you can experiment with different thought patterns. More over, just because it is based on humans doesn't mean it actually has to develop like humans as we know them. If put in it's own world for extended periods of time with others, very different traits and customs could emerge which would help spur new types of technology that perhaps we haven't thought of at all. To think that something like this can be so narrowly categorized is just ignorance, but that's why I'm writing this series anyway.
  4. Death BOO Z's Avatar
    I remember reading about a computer called the "General Learning Machine'. from what I recall, it was loaded with a set of ten axioms and used them to generate more laws and formulas.

    I think A.I depends on consciousness, rather than solving problems.
  5. Kraco's Avatar
    I get the feeling Barumonk's concept of a strong AI is still something completely under human control, even if it's given freedom to simulate something novel. In my opinion something compromised like that would hardly be a strong AI. Merely an average AI at most, to differentiate it from the weak AIs. A real strong AI would be something humans wouldn't necessarily even understand anymore, much less could use it for their own purposes to simulate whatever they wanted. Since an AI could theoretically have unlimited resources, it could easily be thinking a thousand or a million times faster than a human. How would you understand something like that unless it wanted to be understood? Especially since it could completely rewrite itself in the middle of the conversation, appearing an utterly different entity during the time it would take a human to get the next line out of his mouth.
  6. Barumonk's Avatar
    Quote Originally Posted by Kraco
    I get the feeling Barumonk's concept of a strong AI is still something completely under human control, even if it's given freedom to simulate something novel. In my opinion something compromised like that would hardly be a strong AI. Merely an average AI at most, to differentiate it from the weak AIs. A real strong AI would be something humans wouldn't necessarily even understand anymore, much less could use it for their own purposes to simulate whatever they wanted. Since an AI could theoretically have unlimited resources, it could easily be thinking a thousand or a million times faster than a human. How would you understand something like that unless it wanted to be understood? Especially since it could completely rewrite itself in the middle of the conversation, appearing an utterly different entity during the time it would take a human to get the next line out of his mouth.
    I get the feeling that you like to make science fiction seem mythical and impossible to explain. There are a lot of things humans make where we don't fully understand whats going on behind the scenes, but computer programming really isn't like that. Depending on how it's programmed, it could still qualify as a strong AI and be restrained within a command system. The final product doesn't necessarily have to be "Fuck the world! I'm my own entity now, I don't have to do anything you want." Please realize that once we actually make a strong AI, we will understand how it works, why it works, and how to alter it. In fact if the person or team that actually makes the first strong AI doesn't heavily document it for the purposes of patent then I'll eat my underpants with hot sauce on them.

    An extremely simplistic example is having two parts to the system where one decides what it will do, while the other decides how it will do it. This is actually a lot easier to say than it is to program it because it requires abstract thought and the ability to properly evaluate the variables of a given situation while allowing the AI to be flexible, but I digress. A simple command system will bypass the first system to decide what it will do, and it will cause the AI to begin the process by evaluating how it will do it. Further more a priority queue for the products of the second system would allow it to decide when the best time is to carry out a command. So that if it's in a car on the highway and you tell it to mow the lawn later then it won't try to open the door while you're moving to step out of the car, and it will remember to remind you that the portable gas tank for the mower is empty so we need to stop by a gas station on the way back.

    Quote Originally Posted by Kraco
    Especially since it could completely rewrite itself in the middle of the conversation, appearing an utterly different entity during the time it would take a human to get the next line out of his mouth.
    Reflection in programming is pretty high level meta that is usually only fully implemented by virtual machine languages. I wouldn't rule out a VM but it will probably not implement the entire AI itself, rather it would assist the foundation by allowing for hardware drivers or additional software addons to the foundation. It wouldn't actually be able to alter the foundation itself. This is a solid concept because if the drivers or script crashes, the foundation can reboot them without sacrificing the entire conciousness - this is how a whole category of operating systems work that predate Linux called Microkernels.

    You give AI too much credit, remember that they will be built on the technology of humans.
    Updated Sat, 10-29-2011 at 12:18 PM by Barumonk (Grammar.)
  7. Animeniax's Avatar
    You speak of AI like it's a computer that's only able to carry out what functions we've programmed it to. The very definition of AI is that it will have the power of computation on top of human creativity and ingenuity. It will be able to form patterns and calculations from existing information to create new methods and discoveries that humans cannot because of the mess of emotion and ego that go with human thought along with the lack of computing power of the human brain. With an AI, it just has to shut off its emotion chip and then it can exceed any human capacity.
  8. Kraco's Avatar
    After that reply, I'm all the more convinced Baru isn't talking about a real AI but something achievable in the next decades. Something you could entrust piloting your car to but little else. Perhaps even commandeering a surgical robot. Very, very contemporary practicality based thinking without an ounce of imagination. I suppose that's okay, since lines like "if the person or team that actually makes the first strong AI doesn't heavily document it for the purposes of patent" tell all about his priorities. It's the same ludicrous thinking that makes self-proclaimed researchers apply for patents of human genes and self-proclaimed civil servants accepting those applications.
  9. Barumonk's Avatar
    Quote Originally Posted by Animeniax
    You speak of AI like it's a computer that's only able to carry out what functions we've programmed it to.
    If you actually read everything so far, you would know that I was speaking of how others are making AI and how I think that's wrong. Please refer to the sentence in the first part "Traditionally AI programmers have thought that if you build enough weak AI systems and string them together you get a strong AI, and I'm here to tell you that they are fucking retarded and don't deserve their job."

    Quote Originally Posted by Kraco
    After that reply, I'm all the more convinced Baru isn't talking about a real AI but something achievable in the next decades. Something you could entrust piloting your car to but little else. Perhaps even commandeering a surgical robot. Very, very contemporary practicality based thinking without an ounce of imagination. I suppose that's okay, since lines like "if the person or team that actually makes the first strong AI doesn't heavily document it for the purposes of patent" tell all about his priorities. It's the same ludicrous thinking that makes self-proclaimed researchers apply for patents of human genes and self-proclaimed civil servants accepting those applications.
    I feel like you're relentlessly looking for some reason to make this impossible. AI comes in many forms, and regardless of your personal beliefs there are actual definitions in the industry and that is exactly what I'm using. Your idea of a real AI is only yours and not shared by the people who actually write them. More over, you have the same problem as the guy I just quoted above. I'm referring to someone else's work which I don't agree with and not my own.

    As I stated at the very beginning, "It will touch a bit on all parts and hopefully give you some insight to the field and where it's going as we are about to head into 2012." This isn't yet about what I think a strong AI is or how it should be built, I'm writing about the most likely places that the technology will emerge from which will make AI practical in everyday life. Yes, I have my own concept of how strong AI should be built, and no I haven't really gotten into it yet. The only hint I've given so far is in my last reply and in Part 3 on Microkernels.
  10. Kraco's Avatar
    Yeah, I'm sorry I appeared so hostile. It was largely due to the fact you are indeed writing about very likely near future AIs. I'm not in the field of trying to create an artificial intelligence of any degree, so my concept of it is a bit more... romanticized than an adaptive controller of a logistical system. But since you were in fact writing about those, it was my bad to attack you based on something you weren't even intending to address. So, yeah, I was off-topic. I'll leave you in peace now.