[ExI] What SF do you plan to read next?

[Damien Broderick]

On 3/27/2010 2:05 PM, BillK quoth:

> The robots in Sladek's
> story all manage to find logical loopholes in the Three Laws, usually
> with bloody results. Sladek later wrote a novel, Tik-Tok (1983), in
> which a robot discovers that his so-called "asimov circuits" are not
> restraining his behavior at all, making him in effect a sociopath; he
> comes to doubt whether "asimov circuits" are even technically
> possible, deciding that they are simply a pseudo-religious belief held
> by robots.


Indeed. In my novella "The Ballad of Bowsprit Bear's Stead" (hint: 
reprinted in UNCLE BONES), my robots Marx and Smith describe their own 
Three Laws:

========

"Tell me, are you robots hard-wired to tell the truth?"
	"Categorically," Marx affirmed stoutly.
	"If you'll forgive me, sir," Smith added, "that was a rather pointless 
exchange. You've run up against the paradox of the Cretan Liar, sir. If 
Marx is a liar, how can you trust a word he says?"
	"Quite so." I cudgeled my brains. Stepping closer, I discerned a line 
of print stamped into their looming hulls, one in English, one in 
Mandarin. Illiterate in either, I asked Roger, "Is that the statutory 
warrant that these robots are programmed to obey the Three Laws? Answer 
yes or no."
	"Yes."
	With a note of resonant ritual, Marx said: "We avow our adherence to 
the Three Laws of Microprocessors."
	"First," said Smith, " 'Thou shalt love mankind with thy whole mind and 
thy whole heart and thy whole soul.' "
	"Second," cried Marx, " 'Thou shalt love thy neighbor as thyself.' "
	"Third," finished Smith, clashing its heels together in a crisp salute, 
" 'Thou shalt love thyself.' "
	I was shaken; I'd imagined the behemoths under the control of a more 
stringent algorithm than that. "It seems rather open to interpretation."
	"Ethics is like that," Marx said. "It's a Gödel problem, like the 
Cretan Liar. Don't fret, though, sir. We're situationalists, but we opt 
from a rather comprehensive metaphysical consensus."

===============

I rather like to think that Eliezer's AIs will follow similar 
situationalist guidance.

Damien Broderick