Rational Morality

Consider the following gems of wisdom:

• The one alienating his friends will soon be alone.
• You will find your destiny – just start looking in the right places.
• Don’t start to cuddle if she likes it rough.

Some might read these sentences and and start pondering them as good advice. “Ahh – I get it now, I don’t find my destiny because I have been looking in the wrong places all this time. Gee – thanks!” or “Oh yes of course – I should be nice to my friends – thanks for reminding me!” Well, gee whiz indeed, Cracker Jack – they are all tautological:

“A rhetorical tautology can be defined as a series of statements that comprise an argument, whereby the statements are constructed in such a way that the truth of the propositions are guaranteed or that the truth of the propositions cannot be disputed by defining a term in terms of another self referentially. Consequently the statement conveys no useful information regardless of its length or complexity making it unfalsifiable. It is a way of formulating a description such that it masquerades as an explanation when the real reason for the phenomena cannot be independently derived.”

For this very reason and in contrast to its first cousins the hendiadys, the pleonasm and the accumulatio, this particular rhetorical device happens to be a logical fallacy when presented in the form of an argument:

“[A] tautology is an argument that utilizes circular reasoning, which means that the conclusion is also its own premise. The structure of such arguments is A=B therefore A=B, although the premise and conclusion might be formulated differently so it is not immediately apparent as such. For example, saying that therapeutic touch works because it manipulates the life force is a tautology because the definition of therapeutic touch is the alleged manipulation (without touching) of the life force.”

Consequently all tautological advice can be summed up in four words: “Do the right thing!” People obviously agree that the right thing should be done, coming to an understanding of what ‘the right thing’ is for any particular situation is where the problem lies and the real work starts.

Let’s examine the example of coherent extrapolated volition (CEV) as a solution to the AI friendliness problem for instance. In summary CEV proposes – and please correct me if I am wrong -  to create a super intelligence and let it solve AI friendliness for us by simulating better versions of ourself and what we then would want as goal for the AI to do:

• what we would decide if we knew more
• thought faster
• were more the people we wished we were
• had grown up farther together

All well and good, right? More knowledge, faster thinking, better people, more mature… But look what kind of advice this would translate into for the AI tasked with CEV:

• Do that which you would do if you knew more.
• Think faster to solve your problems.
• Do what a better person would do.
• Do the mature thing.

The first two items on the list are tautological since obviously a super intelligence would know more and think faster – how else could we be justified in calling it a super intelligence? The last two items on the list are tautological because they contain the conclusion as the premise without specifying content. “Yes – let’s just simulate better / more mature versions of us – great!” No! What is it that defines a better and/or a more  mature person? What is the transcendent standard of value in line of which the super intelligence can measure a better or more mature individual?

Granted, a super intelligence could theoretically simulate a thousand transhuman beings for a million subjective years in a matter of minutes, but that would not help at all – according to this model – not unless it had a reference to measure them against. If you can not measure it you can not build it, because you have no way of knowing when you got it made. And if you can not build it you can not simulate it – no matter how ’super’ you are.

Recognizing this fatal flaw is of course the reason leading to the unfortunate paper clip fallacy…

What puzzles me more than almost anything else about CEV is that the solution to the riddle CEV is trying to solve lies at the very beginning of the very long winded CEV article:

“Designing a framework for an abstract invariant that doesn’t automatically wipe out the human species.“

Here it is starring us right in the face: our existence is preferable over our non-existence. Why not start from there and keep building? And better be quick, tomorrow is just a day away.

This entry was posted on Friday, November 6th, 2009 at 3:53 am and is filed under Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.