 |
Here's a random thought...
It has been shown that if you have a fixed budget, and wish to do some computer-intensive calculation, then if the time for the calculation is *longer* than 4 years, you should not start the calculation, but should wait until the speed of computers has increased such that the calculation will only take 4 years.
If you take 'learning' to be the computational task, the optimal life span of an AI will be 4 years. Which would give the AIs a very short-sighted view of the world, and might be their weakness. Of course - this weakness lasts only as long as Moore's Law holds out - if Moore's Law levels off, then the optimal life span of an AI increases. But maybe it'll give us a second chance at playing dominant species...
What's Moore's Law say?
Athon Solo
Moore's Law is the idea that computer power doubles in speed every two years. It's held roughly true since the 60s...
Four-year lifespan...
Anyone else flashing back to Blade Runner?
we humans don't become obsolete because we don't upgrade our hardware every so often - because we have highly adaptive brains and are able to use and develop tools to help us do what we want.
A decent artificial intelligence ought to have a similarly adaptive mind and, if it needs to do something tricky, surely it would be able to use the latest tools available to assist it.
Please note that Not Panicking Ltd is not responsible for the content of any external sites listed. The content on h2g2 is created by h2g2's Researchers, who are members of the public. Unlike Edited Guide Entries, the content on this page has not necessarily been checked by a h2g2 editor. In the event that you consider anything on this page to be in breach of the site's House Rules, please
click here
.
|
