|
Philosophy Corner: Ontology | by MatthewDBA | 2009-08-24 06:51:21 |
|
A self-evidential event of the subjective ego | by bitflipper | 2009-08-24 07:26:49 |
|
Would be Descartes-esque. | by krikkert | 2009-08-24 07:33:06 |
|
Actually, I was thinking more of conditions that | by bitflipper | 2009-08-24 07:38:07 |
|
As with anything like this, we need a framework... | by jdelphiki | 2009-08-24 08:02:09 |
|
How about "adaptive behavior"? | by bitflipper | 2009-08-24 08:30:22 |
|
That's an interesting approach. | by MatthewDBA | 2009-08-24 08:34:59 |
|
Perhaps it does | by bitflipper | 2009-08-24 09:26:36 |
|
How would you be able to tell | by MatthewDBA | 2009-08-24 09:33:52 |
| No, an external test is sufficient |
by bitflipper |
2009-08-24 09:47:07 |
Run the application a few times to determine to what parameters it is responsive, then begin changing some of those parameters; if the application is not adaptive, it will betray itself quickly. If it is adaptive, an "expert user" will come to suspect so after a few iterations of "what happens when I tweak this parameter, thusly?"
Now, does this mean we can know (without looking at the code, that is) that an AI is truly adaptive? No, not really; we must hold out the possibility that we are being tricked--that it is "only" a sophisticated progam with no situational adaptability whatsoever.
However, the same can be said about gravity and sunrises; until they fail to happen as theory predicts they will, we cannot know that out theories about them are utterly correct and that they will never fail to behave as those theories predict.
But, there does come a point after which we can be very confident in saying "a program this small (as determined by how quickly it loads and responds compared to other applications) and yet this versatile must be adaptive in its algorithms in order to behave thusly." Just as with physical phenomena, we can say with great confidence, this model explains the aspects and bases of this phenomenon well, and so we can predict this phenomenon according to this model--the sun will rise tomorrow, and dropped apples will fall to the ground." |
|
[ Reply ] |
|
I'm not as sure as you seem to be | by MatthewDBA | 2009-08-24 09:59:26 |
|
Most programs are written deliberately to assist | by bitflipper | 2009-08-24 10:47:04 |
|
You can determine whether another person | by MatthewDBA | 2009-08-24 10:51:07 |
|
How is it absent? | by bitflipper | 2009-08-24 11:16:26 |
|
I'm wondering if things like | by MatthewDBA | 2009-08-24 11:28:16 |
|
But, again, would a conversation with a strong AI | by bitflipper | 2009-08-24 11:52:24 |
|
One difference between | by MatthewDBA | 2009-08-24 12:28:51 |
|
Yes, but, by market statistics, most people | by bitflipper | 2009-08-24 13:11:20 |
