[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Date Index][
Thread Index][
Author Index]
Re: 2045: The Year Man Becomes Immortal
Ok, here's my reply after doing a small bit of reading.
To IBM's Watson, it seems to use a "best of" approach to multiple binary search tree algorithms. Maybe I'm wrong, I'm basing this on Wikipedia's entry on Watson, but it had 4Tb of data stored to search from, far more than the data any human would store, and the algorithms it used to search that data are decided non human. A human would not need multiple algorithms, nor would it need to follow each of the different paths to its end in order to select a result. A human would choose a best path based on previous experience and I dare I say "intuition." My argument would be that a "thought" algorithm that simulates humans is still elusive.
I'd argue that the reason IBM's Watson needs so much data and multiple search algorithms is that the architecture of the IBM machine is fundamentally different from that of the human brain.
If the argument is whether or not a machine can be constructed that plays Jeopardy or Chess, that's different. The "deep search" that IBM uses for those objectives is effective, but it is decidedly specific to the task at hand and non human.
I'm not arguing against intelligent machines. I'm arguing that they'd need some sort of analog electro-chemical engine instead of the current binary electronic engine.
And to Toby G, yes Penrose's ideas seem to be widely disputed according to Wikipedia. And yes, I still haven't read Kurzweil's work.
I just feel the proof is in the pudding. If machines (given their current architecture) could be made to think, we'd see thinking machines by now.
On Mon, Feb 21, 2011 at 8:04 PM, Art Simon
<simart@gmail.com> wrote:
Togy G wrote:
Penrose's thesis is widely rejected these
days.
And to Rick, no I haven't seen Watson or read Kurzwiel's "spiritual age of machines."
Let me do a little more reading, and I'll get back to this.
I'm skeptical, but I'll try to keep an open mind.
--
Art Simon
simart@gmail.commyspace [dot] com/artsimon
