Blank’s First Three Theorems of Artificial Intelligence
Artificial Intelligence has long been an intellectual 'ear-worm' of mine -- something that totally consumes my fancy, and which I can't seem to get out of my head. I've read a lot of texts on the subject (I really recommend the texts by Hans Moravec, if you're looking) and in college I wrote a term paper for Prof. Robert Thurman on the implications of AI from a buddhist perspective. (Long story short, you are far better off assuming AI is real and being wrong, than assuming AI isn't real and being wrong -- it's just like being pregnant).
Over the past couple of years a few thoughts along these lines have been fluttering in the back of my head, and I wanted to take this opportunity to share them.
Blank's First Theorem of Artificial Intelligence states that: Artificial Intelligence is Inevitable. It is not a random pipe-dream or something that will only possibly occur -- it is the inevitable result of the trajectory on which we are headed.
I believe this by fact of:
- the exponentially increasing power of networks, constituted from an exponentially increasing number of nodes (hardware and software), each of which is increasing in power at an exponential rate (computers are always more powerful, there are always more of them, and they are all connected)
- the fact that life permeates every thing it can
Alan Turing postulated what is still considered the most rigorous test for the presence of artificial intelligence. Paraphrased, the Turing Test states that 'if an entity can convince you that it is intelligent, it is.' But, if we look at this from a different perspective, if an entity can not convince you it is intelligent, does that mean it is not?
To me, this is obviously untrue. I am still shocked at the number of people who deny that dogs are intelligent. To me, it is blatantly obvious that every living thing exhibits something we would call intelligence. To others, this is not only non-obvious -- it's insane.
This leads to Blank's Second Theorem of Artificial Intelligence: if an entity believes it is intelligent, you must treat it as though it is. (Whether or not an entity is actually intelligent is irrelevant to the discussion).
This is an important change from the Turing Test. Turing placed the onus of 'proof of intelligence' on the act of convincing another. I am placing the onus of 'proof of intelligence' on one's own self. If I believe I am intelligent, who are you to say that I am not?
The tension implicit in that last question leads directly to Blank's Third Theorem of Artificial Intelligence: The failure to adopt the Second Theorem will inevitably lead to violence between those that consider themselves intelligent, and those others whom they do not believe are intelligent. (See 'Matrix, The' -- but for heaven's sake, skip the sequels!).
If an entity can not convince you it is intelligent, how are you going to convince it that it is not? In other words, the question of 'are you really intelligent or not?' (or, more broadly, 'is Artificial Intelligence real?') is not a discussion worth having.
Before signing off, I'd like to throw in a fourth theorem -- not directly tied to the other three, but one which I think provides some urgency to these ideas. Blank's Fourth Theorem of Artificial Intelligence, states that: when, at some point in the future, the existence of AI becomes incontrovertible, society will estimate a date at which AI was born; that date will precede today.
Shortly stated: AI already exists.
In summary, to restate the Theorems:
First: Artificial Intelligence is inevitable
Second: if an entity believes it is intelligent, you must treat it as though it is
Third: otherwise, violence will ensue
Fourth: Artificial Intelligence already exists
Share and enjoy!
-r
6 comments
It’s worth adding that ‘treating an entity as though it is intelligent’ is a different thing from ‘agreeing that the other entity is intelligent’. The point is this is about individual respect more than a rigorous scientific definition.
Social comments and analytics for this post…
This post was mentioned on Twitter by rblank9: Got time to waste? Blank’s First Three Theorems of Artificial Intelligence: http://is.gd/4DhKG...
[WORDPRESS HASHCASH] The comment’s server IP (75.101.186.211) doesn’t match the comment’s URL host IP (72.47.254.135) and so is spam.
What are your thoughts on intelligence requiring consciousness and/or awareness? Some AI experts have now divided AI into different fields and called one of them “computational intelligence”. This is an interesting division to make, since it suggests they believe there are other forms of intelligence that are not computational and this also ties in with discussions on whether thought is computational or not.
The fourth one. I like it.
Also, I wonder to what extent other intelligences consider humans lacking in intelligence?
@Gerard Toonstra
I don’t really think about this stuff enough to use these sorts of labels. But they are interesting. I suppose the opposite of ‘computational’ intelligence would be ‘emotional’ intelligence? In which case, I do see a use for this separation. It seems to me to roughly correspond to the left-brain/right-brain dichotomy.
So, if I had to make a more structured definition of ‘computational intelligence’ as opposed to ‘emotional intelligence’, it would be this: ‘computational’ intelligence tells you that I’m wearing a red plaid shirt; ‘emotional’ intelligence is what would tell you that the shirt is hideous. In essence, I would argue that computational intelligence involves pattern detection (that’s red plaid), and emotional intelligence requires the cross-modal translation of those patterns (red plaid is ugly and makes me think of my great aunt Martha, who smelled bad, and thus this shade of red smells bad).
I think it’s inarguable that we have achieved artificial computational intelligence in any number of realms. The ‘emotional’ intelligence is definitely a cooler trick to pull off (though, again, I think we’re there already).
I’m actually trying to work on a bit of sample Flash code right now, to accompany a talk I’m writing for 2010. This code represents an attempt to execute some simple emotional intelligence by executing cross-modal translations. So I s’pose, stay tuned!
And yes, I do believe that intelligence implies awareness (and I do believe that awareness and consciousness are essentially synonymous), and that both exist in degrees. An entity could not be said to be either intelligent/aware or not; instead an entity’s intelligence/awareness could be said to be measured on a scale from diminishing to increasing intelligence/awareness.
@Cortexelus
- The fourth one. I like it.
Thank you. The post actually started with the fourth one. But then I figured it needed an intro. Hence the first three. So yes, 75% of my theorems are filler
- Also, I wonder to what extent other intelligences consider humans lacking in intelligence?
Exactly my point.