## Thursday, December 4, 2014

### In honor of Our Future Robot Overlords, let’s conduct a thought experiment: What’s believable?

Let's arrange some propositions along a scale from -10, totally unbelievable, to +10, totally believable.

For example:
(1) The sun will come up tomorrow. +10
(2) Barack Obama is President of the United States. +10
(3) The moon is made of green cheese. -10
(4) Dogs are man’s best friend. +6
(5) Cats a better pets than dogs. +1
(6) MacDonald’s serves good healthy food. -3
(7) Two plus two is four. +10
(8) Seven times nine is thirty-eight. -10
(9) Hoboken is a better place to live than Jersey City. 0
Now, I’m not going to insist that my ratings of those propositions are universally true. You may or may not agree with them. You might also suspect that this is not an entirely rational way to proceed. I respect that.

This is not a profound exercise. It’s a game. And those ratings give you some idea of how I think about things. Think of them as calibration.

Now we’re going get just a little more serious.
(10) The second coming of Christ is inevitable, but we can’t predict when that will happen. -9
There are, of course, people who find that statement totally believable. Obviously I’m not one of them. You might wonder, however, why I didn’t run my rating on that one all the way out there to -10. Well, if I’d done that, then what do I do with this?
(11) The second coming of Christ is going to happen in three hours (it’s now 9AM DST in Jersey City, NJ).
It seems to me that that statement is even less believable than 10. I had to award statement 10 a -9 so I’d have room to award -10 to statement 11.

Ah, but what about statement 12, isn’t it even less believable than 11?
(12) The second coming of Christ happened on January 29, 2009 and we’re now all either in Heaven, Limbo, or Hell.
If statement 11 is -10, don’t we have to rate number 12 as off the scale -11? This is a tricky little game we’re playing.

Now we get to the good part:
(13) Belief in the likelihood of superintelligent machines makes as much sense as belief in the second coming of Christ.
Notice that I’ve not assigned a level of belief to that statement. Notice also that unlike statements 1 through 12, statement 13 is about level of belief. It is asserting that one statement – something like “superintelligent machines are going to happen” – has much the same belief status as another statement – “Christ will be born a second time.” But I’ve presented three different versions of that second statement – numbers 10, 11, and 12 – to which I’ve assigned different levels of belief.

And that’s my point.
(14) I think that belief in superintelligent machines is unwarranted and that it is unwarranted in degrees similar to belief in the second coming of Christ.
Now it gets really tricky. I don’t quite believe 14.

Belief in the second coming of Christ has been around for a long time. And I’m sure that there are those who will affirm their belief in the Second Coming but who will also tell that theirs is not a literal belief. It’s some other kind of belief, to be judged by some other standard. I think that qualification must be taken seriously.

But we must also consider the fact that belief in superintelligent machines is quite recent and arises in response to technology that’s only been around for fifty or sixty years. We really don’t know what’s going to come of that technology in the next 100 years. I certainly don’t.

Those who believe in the likely coming of superintelligent machines have no technical arguments to offer. Their arguments are different in kind, for example, from arguments about the feasibility of sending a manned mission to Mars. Further, the short history of AU is littered with undelivered promises of future intelligence. It’s a history that doesn’t engender confidence in AI’s ability to predict future developments. Finally, while many of those who believe in superintelligent machines are technically sophisticated, many doubters are also technically sophisticated. On that score it’s a standoff.

On the matter of machine intelligence we simply don’t know. But, as I said, the issue hasn’t been around for very long. We’ll know more a hundred years from now. At that time, how will people score the following proposition?
(15) Belief in superintelligent machines is to be judged in the same way as belief in the second coming of Christ.