Author Topic: robots replacing humans (thought provoking post)  (Read 8944 times)

0 Members and 1 Guest are viewing this topic.

Offline AdminTopic starter

  • Administrator
  • Supreme Robot
  • *****
  • Posts: 11,703
  • Helpful? 173
    • Society of Robots
robots replacing humans (thought provoking post)
« on: September 02, 2006, 05:49:21 PM »
When do you think a robot capable of human level thought, or mobility, will be created? What advances that need to be made before this can happen?

Offline dunk

  • Expert Roboticist
  • Supreme Robot
  • *****
  • Posts: 1,086
  • Helpful? 21
    • dunk's robot
Re: robots replacing humans (thought provoking post)
« Reply #1 on: September 03, 2006, 08:19:42 AM »
i think the question on mobility is the simplest to answer. with modern sensors and materials it is allready possible to build machines that can traverse and interact with the environment in a similar way to people. the technology is allready there it just needs refining.
machines can allready outdo humans on this level in some environments. underwater robots and flying robots would be 2 examples of robots allready being more at home in an environment than humans.

the other half of your question depends on what you mean by "thought".
if you are talking about the ability to make decisions and act upon them without background knowledge of the problem then you are comparing machine intelligence to humans problem solving ability.

in the present, machines have no real level of problem solving techniques. any "Artificial Intelligence" algorthm currently in use must be specifically tailered to it's application.
machines will not be self determining until someone invents an algorithm that allows machines to select what problem to solve rather than just solving what they are told to.
as to when this invention might occur, it might be tomorrow, it might be never. it may be some simple combination of mathematics and logic or it may be a huge cumbersome programming challenge.
that's the problem with inventions. it's hard to predict what you're going to get.

if on the other hand you rate "human level thought" as conciseness then it complicates things somewhat.
first you would need to debate what is it to be conscious. are we conscious or do we just react to stimulus provided by our surroundings? are we really just complicated machines or do we indeed have something extra?

let us presume for a moment that we do indeed have that "extra" ingredient that makes us conscious.
if you were to create this in a machine you would by definition made it something more than a machine.
this is a philosophical minefield right here. tread lightly.

i would imagine any conscious machine would need a degree of problem solving ability but whether a conscious machine would do all it's thought processing as part of one "consciousness" algorithm, or built in layers, with "problem solving" operating on one layer and "consciousness" in a separate sub program i could not say.
as with the "machine problem solving" argument, such a break through might occur tomorrow in the future or never. it needs an invention to occur.
in my opinion we can expect a breakthrough in machine problem solving long before we can expect one in machine consciousness.


if such breakthroughs occurred i think machine intelligence would progress very rapidly. look at the rate of progress in the IT industry in the last 20 years. if it would be fair to say that modern computer double their processing power every 2 years or so, i think we could expect the same from machine intelligence.

dunk.

Offline AdminTopic starter

  • Administrator
  • Supreme Robot
  • *****
  • Posts: 11,703
  • Helpful? 173
    • Society of Robots
Re: robots replacing humans (thought provoking post)
« Reply #2 on: September 03, 2006, 08:37:17 AM »
Ok I guess what I meant by human level thought is that in terms of functionality it can accomplish similar goals. The same robot can play baseball one moment, then go pack a suitcase the next moment. The actually mechanism in the robot brain can be anything (including some super computer transmitting wirelessly), just as long as the robot is comparable to a human in function (can do many unrelated tasks and operate in a human environment effectively).

As for mobility, I guess this was more open ended. What I originally thought was a robot that can lift boxes then go rockclimbing afterwards. But I guess a robot that can hover in a house or up stairs would be just as effective . . .

Offline Gopher

  • Full Member
  • ***
  • Posts: 82
  • Helpful? 0
Re: robots replacing humans (thought provoking post)
« Reply #3 on: October 05, 2006, 05:56:54 PM »
The subject of consciousness is a difficult one for philosophical reasons. Many of a more spiritual bend than I think that consciousness is something beyond the physical, related to some transcendent concept such as the soul. I don't believe this myself, and to be honest I don't the concept of consciousness is really well-defined enough to even debate resonably at present. To me, consciousness implies simply awareness of yourself and your environment. I see this as a prerequisite for anything beyond the most rudimentary, instinctive behavior. By this standard, a chimpanzee's ability to learn by immitation behaviors like termite fishing or breaking open coconuts qualifies it as having "conscious." Some think imitation is instinctive, but I'm not sure I agree. It implies an awareness of the other chimpanzee, and an ability to recognize, if not understand, causal relationships. Interestingly, I think this is a larger accomplishment than figuring out how to bash the coconut open in the first place - that can happen by accident, or out of desperation. Imitating the action when witnessed implies an ability to watch another, and imagine yourself in that position, basically think "That could be me, I could be eating that."

As far as I know, acheiving this level of behavior in a robot is far beyond our present ability, though as much because of the difficulty of processing visual data in so sophisticated a manner as to identify not only another robot but recognize an action taken by that robot that is not part of the observer robot's programming.

There are other, probably better, ways to measure intelligence, but in my view we will not have true human-level intelligence in a robot until that robot is capable of abstract thought - not necessarily in human language, but in some abstracted language that serves the same purpose for the machine. Without that, the best we can hope for is the robotic equivilant of a dog or cat, capable of some degree of adaptability but not capable of really understanding or consciously impacting it's environment. Without the ability to reason in a general, abstracted way (as opposed to the specific, restricted way that expert machines reason) a robot can't reach a human-level of sophistication.

As for the mobility question, I think your original thought was humanoid mobility, rather than just mobility in general? And as far as I am concerned, a real human level of kinematic intelligence requires some of that same reasoning ability above; you might program a bot to play baseball, or climb rocks, but no matter how many specific actions it is programmed with, it is no more likely to learn an entirely new behavior on it's own - say, climbing a rope ladder - than any other humanoid bot. To make that leap requires an understanding of your body and the ability to apply that knowledge to new tasks. When we can make a robot that can learn a sport or martial art the same way a human does, through practice, observation, and verbal instruction rather than through programming or mechanical imitation, then we will have a truly human level of intelligence and mobility.

I hope we'll accomplish this goal in the next 50 years, but I'm not knowledgable enough about the subject to really guess with any accuracy.

 


data_list