My World View – Intelligence

Yes, my ‘world view’ seems to cover more than gods and morality. I think everyone’s world view does but they seem to forget that most of the time. Recently I’ve seen a few articles and posts talking about Artificial Intelligence. This is something that I’m quite interested in and have been since I was very little. It does not directly tell us how we think and remember or why, but it’s an important step. Here is a post about the brief history of AI and Hollywood is not involved. For those that have been reading along, it might be of interest to know that Alan Turing was homosexual. How poor our species would be if we ignored some section of it based only on sexual orientation.

Moving on

AI has been thought of as trying to imitate the intelligence of the Human ape. I really REALLY do not think this is a good place to start. It’s akin to teaching art by telling students to imitate Renoir. I think a much more sensible place to start is to try to imitate something as technically difficult as Dr Suess’s illustrations, or more correctly something on this list that has a low neuron count.

What do you mean?

Intelligence is often confused by people who think that it necessarily involves philosophy. Trust me, reading what philosophers put out I often find it difficult to believe that any intelligence was involved at all. So what is intelligence?

Wikipedia says that…  Numerous definitions of and hypotheses about intelligence have been proposed since before the twentieth century, with no consensus reached by scholars. This  should tell you something important … that reductionists have not yet had a loud voice in this discussion. If you take a reductionist view, intelligence can be said to be the smallest ability required to act in reaction to external stimuli. Yes, by this definition a roundworm is an intelligent life form. We generally ascribe intelligent behavior along a gradient and in direct relationship to the number of neurons and brain size of the animal, though this does not hold exactingly true in all cases.

We should start with 12 or 15 neurons in our AI development systems and find out what we can possibly make out of that. This is how evolution did it… it should work for us too. I am still looking for experiments that work this way – bottom up rather than top down design. Everyone reading this probably has an example of such a system in their home. The lowly thermostat which controls the temperature in your house is such a system. A simple switch that reacts to temperature as an external input.

Really, how does this define intelligence?

If we can see the thermostat as an intelligent system, the definition of intelligence rests somewhere near this: Intelligence is the ability to react to external stimuli. We can then state that intelligence without changing input is static or more pointedly, it is useless. Evolution did not create intelligence because it’s a nice feature, but because it is useful. In fact it is useful enough to warrant spending great amounts of energy on it. Clearly evolution started with a small number of neurons and nerves and worked it’s way up. Intelligence starts small and gets increasingly larger and more competent over time in the natural world. If we cannot define AI in this way, we’ll be continuously trying to imitate Renoir instead of building what might be better AI systems than can be derived from imitating what is not yet understood… and that is the intelligent way to look at the problem.

How do we get from 100 neurons to trillions? Slowly. Now that we have them, has the definition changed? No. My friend Don will tell us that now we can think about how and what we think and how and why we make decisions where the lowly roundworm can only seem to react to external stimuli. This seems empirically true, but I don’t think that it changes the intrinsic definition of intelligence, only that the extra ability allows for decisions which involved both real stimuli and meta information about the stimuli. Where we cannot tell what an object is, we can guess what it might be by the properties of it that we can discern. This is a skill which many smart people ‘guess’ developed because long ago, when you’re interested in not becoming a meal, you have to learn the difference between a small mammal rustling through the forest floor and a big hungry cat walking across the forest floor. Our pattern recognition had to get much better to survive. More pattern recognition gives us more information with which to make a decision. Then we later would be able to think about patterns within patterns and so on. Human intelligence has many layers built on top of each other and all built on top of the same systems that roundworms have.

This is where we should start developing AI, a couple dozen neurons. It’s useful to keep the philosophy layers out of it while we’re working out how to kind of put it all together on silicon. Likewise, it would be useful to try removing philosophy from human thought to look at one of the ways that advanced intelligent machines could work. Not an easy trick, but someone has to work on it.

Maybe this leaves us with the conclusion that we should see wrist bands which say WWARD or what would a roundworm do? Clearly all our human thinking and rethinking is not necessary for either intelligent action or life. It’s probably causing more problems for us than it solves these days. These are the reasons that I think AI research is going in the wrong direction – they have not defined intelligence correctly. Further, calling the way human apes think the pinnacle of intelligence is, in all probability, hubris.

Advertisements
  1. Dearest Mal,

    First and foremost as soon as I read over your mentioning of AI…my eyes rolled. You do this to me.

    Oh god! More?! A link to the history of AI?…This should be an interesting post.

    I’m listening….

    So intelligence is not consciousness? Do you not need a consciousness to be intellectually correct? Is intellectually correct a correct term? Well yes, when you are speaking consciously. 😉

    “Intelligence is the ability to react to external stimuli. We can then state that intelligence without changing input is static or more pointedly, it is useless.”
    Wonderful words. Bravo.
    The level of intelligence for a human is defined by how well they react to external stimuli. Yes, so right. What is the external stimuli for a gamer? If they consist of a certain intelligence how can they even view a higher, wider, scarier, stimuli?

    You sit outside and look in…their “external” stimuli is actually internal!?!? What do you do now? I bet you feel pretty fucking smart huh? Nah we are actually really fucking normal…but our consciousness is out the roof.

    So…consciousness = intelligence?

    If no input is changing, if the intelligence become static from no external stimuli…this means no conscious…futile and useless. The entire gamer race?

    Hmmm…

    ~bri

    • FUCK, you had to go and stir the pot. Now I have to think this over again. bri, you do this to me and I appreciate you immeasurably for this. I’ll have to get back to you because this will take some thinking. That put a real kink in that line of thinking. Thank you

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: