1. Joined
    06 Mar '12
    Moves
    625
    25 Sep '18 19:186 edits
    https://techxplore.com/news/2018-09-robotic-lettuce-leaf-kind.html

    This one looks like a formidable AI challenge and not sure I would like to have a go at it. My guess is that they won't be able to develop an AI robot able to do a job like this nearly as good as a human until they at least develop a robotic hand that roughly looks like a human hand with a compressible soft-surface (so not to damage soft delicate things it touches) and with a very pressure sensitive sense of touch as good as that of a human hand. But even that would only solve one part of the very big problem with the other big part of the problem being how the AI will handle such extremely irregular AI-unfrendly sensory data. I know from my expertise that using neural network computing for dealing with complex problems like this (which would be the 'obvious' solution to an AI expert) is always far far easier said than done.
  2. Subscribersonhouse
    Fast and Curious
    slatington, pa, usa
    Joined
    28 Dec '04
    Moves
    52617
    25 Sep '18 20:27
    Originally posted by @humy
    https://techxplore.com/news/2018-09-robotic-lettuce-leaf-kind.html

    This one looks like a formidable AI challenge and not sure I would like to have a go at it. My guess is that they won't be able to develop an AI robot able to do a job like this nearly as good as a human until they at least develop a robotic hand that roughly looks like a human hand with a c ...[text shortened]... (which would be the 'obvious' solution to an AI expert) is always far far easier said than done.
    I wonder what that will ultimately do to the farm labor situation? will they get pushed out by automation? Picking cotton for instance. If a robot could do that job, there would be much less to deal with in terms of cuts all cotton pickers endure. Or picking fruit, there is a problme right now where farmers are hurting for labor since Mexican immigrants are staying away, even before Trump's barbarous border policies. So crops are rotting on the vine with less labor available to pick. Of course that is a long term issue, if say, apple pickers get replaced by machines, it won't be for a decade or more so there will be not much effect on farmers needing laborers or laborers let go because machines can do it faster and cheaper with zero worker injuries to worry about.
  3. SubscriberAThousandYoung
    Just another day
    tinyurl.com/y8wgt7a5
    Joined
    23 Aug '04
    Moves
    24791
    25 Sep '18 21:49
    Originally posted by @sonhouse
    I wonder what that will ultimately do to the farm labor situation? will they get pushed out by automation? Picking cotton for instance. If a robot could do that job, there would be much less to deal with in terms of cuts all cotton pickers endure. Or picking fruit, there is a problme right now where farmers are hurting for labor since Mexican immigrants ar ...[text shortened]... s let go because machines can do it faster and cheaper with zero worker injuries to worry about.
    The cotton gin made slave labor MORE valuable, not less
  4. Standard memberchaney3
    Bring it on.......
    Joined
    26 Dec '14
    Moves
    29697
    25 Sep '18 23:23
    Originally posted by @humy
    https://techxplore.com/news/2018-09-robotic-lettuce-leaf-kind.html

    This one looks like a formidable AI challenge and not sure I would like to have a go at it. My guess is that they won't be able to develop an AI robot able to do a job like this nearly as good as a human until they at least develop a robotic hand that roughly looks like a human hand with a c ...[text shortened]... (which would be the 'obvious' solution to an AI expert) is always far far easier said than done.
    It's amazing that all of science and money cannot produce anything that operates even close to what an average 4 year old can do naturally, and with ease.

    Makes "creation" look more and more likely.
  5. Subscribersonhouse
    Fast and Curious
    slatington, pa, usa
    Joined
    28 Dec '04
    Moves
    52617
    25 Sep '18 23:461 edit
    Originally posted by @chaney3
    It's amazing that all of science and money cannot produce anything that operates even close to what an average 4 year old can do naturally, and with ease.

    Makes "creation" look more and more likely.
    Oh come on, science if you didn't already know, is still in kindergarten, it's only a couple hundred years old. You don't have much time left before science is in first grade and probably will have the ability to make robots with the capability of some human functions, I don't imagine robots are going to be able to compete with Itzhah Perlman any time soon but picking cotton or doing lettuce might be within AI capabilities in a few years.
  6. Standard memberchaney3
    Bring it on.......
    Joined
    26 Dec '14
    Moves
    29697
    25 Sep '18 23:58
    Originally posted by @sonhouse
    Oh come on, science if you didn't already know, is still in kindergarten, it's only a couple hundred years old. You don't have much time left before science is in first grade and probably will have the ability to make robots with the capability of some human functions, I don't imagine robots are going to be able to compete with Itzhah Perlman any time soon but picking cotton or doing lettuce might be within AI capabilities in a few years.
    I watched a show on the Discovery channel a few years ago, where they took top scientists and gave them millions of dollars to "create" a robot.

    The thing was a hunk of garbage, which also couldn't do what a 4 year old can.

    The funniest part was having the genius scientists explain how they didn't even attempt to put a digestive system into their robot, because they had no clue how to duplicate it, at all.

    Face it, when a crawling infant can outmaneuver science's best robot, then "real creation" should be seriously addressed by the kindergarten scientists.
  7. Subscribersonhouse
    Fast and Curious
    slatington, pa, usa
    Joined
    28 Dec '04
    Moves
    52617
    26 Sep '18 00:331 edit
    Originally posted by @chaney3
    I watched a show on the Discovery channel a few years ago, where they took top scientists and gave them millions of dollars to "create" a robot.

    The thing was a hunk of garbage, which also couldn't do what a 4 year old can.

    The funniest part was having the genius scientists explain how they didn't even attempt to put a digestive system into their robo ...[text shortened]... s best robot, then "real creation" should be seriously addressed by the kindergarten scientists.
    I guess you totally ignored what I just said. Tell me, how long do you think robotics has been around? Since Robby the Robot? You know, the one with the dude inside? R2D2? same thing.

    You know how they tout the driverless car? AI is SO powerful it can take over driving duties and the human can go to sleep? Did you know with all that powerful AI running things 2 people have died because the AI wasn't good enough?

    Does that give you a clue as to what the level of robotics is at?

    Forget a 4 yo kid, all that AI is not good enough to duplicate a frigging BEE much less a human.

    Do you start to get it now?
  8. Joined
    11 Nov '05
    Moves
    43938
    26 Sep '18 16:561 edit
    Originally posted by @humy
    AI robotic lettuce leaf peeling being developed
    I'm not impressed.

    If the robot is shown a lettuce, I would be quite impressed if the robot asks: "And what do you want me to do with this?"
    That would be real AI.
  9. Subscribersonhouse
    Fast and Curious
    slatington, pa, usa
    Joined
    28 Dec '04
    Moves
    52617
    27 Sep '18 20:171 edit
    Originally posted by @fabianfnas
    I'm not impressed.

    If the robot is shown a lettuce, I would be quite impressed if the robot asks: "And what do you want me to do with this?"
    That would be real AI.
    It may develop into something like that. That is how the very advanced Go playing software was developed, Alpha something or other. Now beating the best human Go players, something nobody thought possible this soon into century 21.

    Already, medical diagnostic software are beating humans in percentage of correct diagnoses and I expect that kind of thing to invade more and more human activity.

    Like I have said before, these robots will not be competing with Andre Segovia any time soon or Picasso but they may come close in another 100 years.
  10. Joined
    11 Nov '05
    Moves
    43938
    28 Sep '18 06:21
    Originally posted by @sonhouse
    It may develop into something like that. That is how the very advanced Go playing software was developed, Alpha something or other. Now beating the best human Go players, something nobody thought possible this soon into century 21.

    Already, medical diagnostic software are beating humans in percentage of correct diagnoses and I expect that kind of thing ...[text shortened]... peting with Andre Segovia any time soon or Picasso but they may come close in another 100 years.
    I often object that todays AI has anything to do with intelligence. It's only clever programming, nothing more.

    There were once days that a decent chess program, winning a school tournament, was considered having artificial intelligence. Alpha Go is the same, only with a higher degree of cleverness in its program.

    When a robot ask me "What is this?" when I show it a new board game, I explain and it says "This I want to learn!", then I will call it artificially intelligent. Not before.
  11. Subscribersonhouse
    Fast and Curious
    slatington, pa, usa
    Joined
    28 Dec '04
    Moves
    52617
    28 Sep '18 15:154 edits
    Originally posted by @fabianfnas
    I often object that todays AI has anything to do with intelligence. It's only clever programming, nothing more.

    There were once days that a decent chess program, winning a school tournament, was considered having artificial intelligence. Alpha Go is the same, only with a higher degree of cleverness in its program.

    When a robot ask me "What is this? ...[text shortened]... n and it says "This I want to learn!", then I will call it artificially intelligent. Not before.
    I don't think anyone is saying a lettuce picking or manipulating robot is intelligent in the human sense, just an extension of abilities now added to the list of attributes such a robot can do when before they could not.

    BTW, Alpha Go is not just incrementally better, it is exponentially better and is in fact duplicating, in an admittedly smaller matrix way, the action of a brain, probably on the order of an ant in complexity but still a significant achievement.

    I think when they get up to a trillion such artificial synapses it will show human level intelligence which of course may be 100 years in the future but maybe a lot closer to our time.

    Just remember, the difference between the brain and an ant is not fundamental. It is in the number of synapses and connections in a larger, much larger actually, matrix of cells in the human brain V the ant brain.

    The fundamental construction of both brains are the same, a bunch of neurons connected together, just a LOT more of those connections in a human brain V an ant.

    And of course my description misses a lot of points but you get the idea.

    If we could somehow hack into an ant's brain and connect a billion of those together we might have a rube Goldberg thing that could do human level things, maybe even play as good as Segovia on Guitar.

    Of course for now that is just wishful thinking but a hundred years from now?

    BTW, I think you knew I was laid off from my last job, but now have had two job interviews, one at a huge company I just learned about because they contacted me, Thor Labs, they make optical coatings for lenses, they are like the company once known as Edmond Scientific which I think is now called Edmond optics. I would be responsible for sputtering tools that does some of the coatings for lenses and other optical components. Thor Labs is the main competitor of Edmond optics.
  12. Joined
    11 Nov '05
    Moves
    43938
    30 Sep '18 10:07
    Originally posted by @sonhouse
    It may develop into something like that. That is how the very advanced Go playing software was developed, Alpha something or other. Now beating the best human Go players, something nobody thought possible this soon into century 21.

    Already, medical diagnostic software are beating humans in percentage of correct diagnoses and I expect that kind of thing ...[text shortened]... peting with Andre Segovia any time soon or Picasso but they may come close in another 100 years.
    Not even Alpha Go shows anything that is comparable with intelligence.

    If, on the other hand, I ask it the meaning of this or that move, and he can explain it to me (or a good Go player) why, and adapt it to my level in the game, then I would be more impressed.

    But Alpha Go isn't anything but a program, nothing more. Remember that chess programs were once called artificial intelligence.

    But I am impressed with Alpha Go as a program! Very impressed!
    But is it intelligent in any way? No, it's not.
  13. Subscribersonhouse
    Fast and Curious
    slatington, pa, usa
    Joined
    28 Dec '04
    Moves
    52617
    30 Sep '18 11:50
    Originally posted by @fabianfnas
    Not even Alpha Go shows anything that is comparable with intelligence.

    If, on the other hand, I ask it the meaning of this or that move, and he can explain it to me (or a good Go player) why, and adapt it to my level in the game, then I would be more impressed.

    But Alpha Go isn't anything but a program, nothing more. Remember that chess programs w ...[text shortened]... ed with Alpha Go as a program! Very impressed!
    But is it intelligent in any way? No, it's not.
    Like I said, it has the computing capabilities of an ant. But multiply that by a billion, you may have something that could compete with humans, maybe come up with new math hypotheses on its own and such but don't hold your breath, that may not happen for a hundred years.

    I think eventually it will happen though. I think its a matter of the size of the neuron matrix they can build.
  14. Joined
    11 Nov '05
    Moves
    43938
    01 Oct '18 08:091 edit
    Originally posted by @sonhouse
    Like I said, it has the computing capabilities of an ant. But multiply that by a billion, you may have something that could compete with humans, maybe come up with new math hypotheses on its own and such but don't hold your breath, that may not happen for a hundred years.

    I think eventually it will happen though. I think its a matter of the size of the neuron matrix they can build.
    Every AI program needs a programmer. A R(eal) I(intelligence) don't.
    A clever program isn't AI per se, and not even close to RI.
    Multiply the cleverness of any program with a billion, then you have another clever program, not an AI program. Cleverer than before, but still not near the Real Intelligence of an infant.

    Do you know DrWatson? This program is called AI. I tried to teach it some Spanish, it failed. I tried to teach a boy 5 years of age, he succeeded. And the boy didn't even need programming.

    What they call AI isn't much more than genetic programming, neural networking, machine learning, big data, and some other methods. Clever methods, but still simple programming.

    Bubble sort isn't AI either, yet you can sort billions of data, way faster than any human can do. But still not AI.

    When you can skip the programming phase, then come back and we can talk about Artificial Intelligence.

    Do you know how many synapses there are in an anthill?
  15. Subscribersonhouse
    Fast and Curious
    slatington, pa, usa
    Joined
    28 Dec '04
    Moves
    52617
    05 Oct '18 14:01
    @fabianfnas said
    Every AI program needs a programmer. A R(eal) I(intelligence) don't.
    A clever program isn't AI per se, and not even close to RI.
    Multiply the cleverness of any program with a billion, then you have another clever program, not an AI program. Cleverer than before, but still not near the Real Intelligence of an infant.

    Do you know DrWatson? This program is called AI. I t ...[text shortened]... we can talk about Artificial Intelligence.

    Do you know how many synapses there are in an anthill?
    This is true, the ant colony has a group intelligence way more than a single ant since they are a unit in terms of work done.
    Still, you don't need real intelligence to run a lettuce leaf peeler or any other farming device. You need to be able to do a specific job with a number of variables like how fast the head of lettuce is coming at the machine, is it really a head of lettuce Vs say a grapefruit rolling down the isle or a small baby. Object recognition has come a LONG way in the last 30 years. I don't think it's THAT big a deal to do that lettuce job. I don't see it needing human level intelligence.
Back to Top