Robot Love

More on Robot Love

Am I right? Consider this: would you enjoy watching a TV show in which contestants competed to solve complicated math equations as quickly as possible? Now, would you be excited to see a computer compete against the humans in this contest? I didn’t think so.

Yes, computers can crunch numbers. In fact, in essence, that’s all they do. The natural language used for the questions in Jeopardy are broken down by the computer into bits and bytes and then processed. Very quickly.

From the computer’s point of view, all of the questions are nothing more than math equations to be solved with speed.

It’s a Binary World
Now this one really bugs me: “KG Blankinship” writes in a letter to the New York Times that “of course we can build machines that exhibit purely random behavior by exploiting quantum mechanics as well”.

But before that he says something even more absurd: “Self-awareness and the ability to adapt creatively can also be programmed into a computer”. The statement is self-contradictory but he hits on a truth: “can be programmed” into a computer. Next, he’ll tell us that a computer can program itself. As if the program that told it to program itself could ever be something that was not, no matter how many steps down the chain, the product of human intervention.

Can a computer’s behavior ever be truly “random”? Or is the appearance of randomness merely the irreducible fact that the human’s have hidden the schedule for the behavior from humans by employing elaborate and obtuse mathematical formulae? Yes, always. And it’s always ultimately math. And the computer is always ultimately binary, which means it can never not be math. And if someone jumps up and shouts “yeah, but sooner or later they will find a way to integrate organic cells…” I say that on that day the organic cells will be self-aware or random, not the computer.


Why does it matter? Because sooner or later someone is going to tell someone else that something is true or must be done and can’t be contradicted because a computer said it was true or must be done. No, the programmer said it was true or must be done. The computer is only doing what it can only do: parrot the input of it’s master.

It occurs to me that some of the people defending the idea that computers can “think” like humans operate under the assumption that the human brain is binary in function, that is, that neurons are all either on or off, with no meaningful in-between state. (I suppose you could also argue that a very, very large number of computer chips could attain a level of virtual analog operation, where there are so many simulated “in-between” states that is operates like a human brain.)

It’s an intriguing line of thought. I don’t believe the human brain is binary in that sense. I believe that human beings are an integrated system in which any particular state of virtually any part of the body has an infinite range of values, which, combined with every other part of the body having an infinite range of values, produces an organism that can never be matched by any device that is, by definition, at its fundamental level, always binary.

To believe that human brains are also binary is to impose a reductionist view of biology onto an organism.  You can only believe it if you choose to see only the binary functions of the organism, and ignoring the organic non-binary aspects of the brain.

After the Performance: AI

There has been a bit of noise this week about the IBM computer that supposedly defeated some of the top human Jeopardy Contestants. I have rarely heard such unmitigated bullshit in the past few years. Consider this:

The computer was allowed to store the IMDB and several encyclopedias including Wiki on it’s hard drives. The human was not even allowed to use Google.

The computer did not express the slightest desire to play the game or win. The IBM programmers did. They cheated by having the IMDB and Wiki with them when they played while the human contestants, of course, did not even have a dictionary.

Some of the observers were dazzled that the computer was able to understand a rhyming word– what animal living in a mountainous region rhymes with “Obama”? They were surprised that the computer had been programmed to “know” that llama rhymes with Obama? You are indeed easily impressed.

The odd thing is that the computer’s performance hasn’t even been all that impressive, even if it was actually a “performance” in any human sense of the word. Apparently, it is offered the question in text rather than verbally. 25 IBM programmers in four years couldn’t do better than that? And why does it get a bye on the verbal questions? Human contestants can’t ask for a print out of the question before they offered verbally to other contestants.

This is a scam.

The bottom line, of course, is that computers can’t “think”. They will never think. All they can do is process data. The data and the processing are constructed by humans. The computer contributes nothing but the illusion of autonomous operation.

People who think computers think are staring at the puppets at a puppet show and wondering what they do at night after the performance.

Robots Can’t Love

I enjoyed “Wall-E”, because the graphics were nice, and the action was wittily contrived. Wall-E meets and falls in love with a more up-to-date computer that looks like an inflated iPod. The two coo for each other.

But why do people so readily want to believe that robots might some day be capable of having feelings?

This is an immutable and irrevocable fact about robots: robots are programs– there is not a single thing they will ever do that is not the result of a programming instruction placed there by a human technician. The “feelings” expressed by a robot will always and ever be as real to anyone as the cuckoo in a cuckoo clock, or those dolls that used to have a string in the back, and will probably be twice as annoying after the very, very brief phase of novelty wears off.

Well, there are movies about talking dogs and flying men and 12-year-olds who know Unix (Jurassic Park), so what’s the problem? The problem is, I get why we might have a compelling movie about a talking dog, or a smart 12-year-old, or a man with superpowers: all of them correspond to real beings who have real feelings, and there are explanations for the dog, the 12-year-old, and superman. There is no explanation than can possibly explain why a robot would have human feelings, just as there is no possible explanation of why a bullet might fly at 10 miles per hour, or there would be a parking spot available right in front of that downtown office building our hero needs to enter immediately.

The problem is, I just don’t find a story line about a robot with feelings compelling. It’s just not interesting. It’s impossible to care about the robot with feelings because I can’t escape the awareness of the fact that every action the robot takes in response to his “feelings” is, in fact, the result of a program created by his manufacturer.

Ironically, the most interesting idea about a robot with “feelings” is this: what if the humans in the story didn’t know it was a robot?


What about “Blade Runner”?

All right– this is an interesting movie. But the “replicants” are clearly not robots– they’re genetically engineered organisms. Or are they? The movie doesn’t explain. They bleed and they die and they have feelings. Does that answer the question? Yes it does– they are organisms, genetically engineered to function like humans, so they can work and live where humans would find conditions intolerable.

But… in one scene, Deckard encounters a maker of the eyes which are clearly manufactured, aren’t they? In fact, they could just as likely have been cultured in some way, grown from stem cells, or what have you.

The most beautiful moment in the movie comes when a replicant does something absolutely human– gets nostalgic:

I’ve seen things you people wouldn’t believe… attack ships on fire off the shoulder of Orion. I watched c-beams glitter in the dark near Tanhauser Gate.

…All those moments will be lost in time, like tears in rain. Time to die.


Other concepts that a Hollywood Producer found credible:

  • perky brain surgeons who look and talk like Meg Ryan
  • parking spaces in front of the building you suddenly need to enter very quickly in order to save someone’s life
  • soldiers as heroes who never seem to actually kill anybody
  • mothers who have all the time in the world to send their children off to school with hugs and kisses and expressions of consuming devotion– as if they knew something bad was going to happen
  • annoying mentally disturbed men who seem strangely attractive to young, beautiful women
  • rogue police who “break all the rules”
  • suspects who immediately tell the truth when threatened by the rogue cop who breaks all the rules.