> Robin Hanson wrote:
> >
> > The fact that people are prone to take these estimate 
> >questions as attitude surveys is all the more reason to seek concrete 
> >arguments, rather than yet more attitudes.

What makes you think that concerete arguments can be found for 
prognostication? 

Yes, Boeing can give you a concrete argument for wat kind of jet
engine it will be building in 10 years (or IBM, about cpu's), 
but just try predicting the atom bomb in 1895.

In 1895, most scientists didn't beleive in atoms, with folks
like Rutherford the exception, not the rule. In 1903, H.G. Wells
wrote a book predicting the atom bomb (and atomic warfare, and
world government). In 1913, Bohr's model of the atom was 
given legitimacy by Mosley's spectroscopy data. Lise Meitner 
and Otto Han didn't figure out that the missing mass was 
just E=mc^2 until the  1930s, and then we had the manhattan 
project.  That's just 50 years from something that didn't 
exist to in-your-face reality. 

I doubt that, even now, you could build a "concrete argument"
for the possibility of the atomic bomb in 1928. In 1928, the
temptation would have been to say either "never, its scientifically
impossible", or to say "gosh, maybe, in thousands of years",
with the latter only because one had some wishy-washy liberal 
optimistic view about the forward progress of science.

The only concrete argument is this: in 1928, we still needed
a few fundamental breakthroughs in physics.

Today, its not clear that we need any fundamental breakthroughs
at all. There's a fair and widespread beleif that simply doing
what we do today, with neural nets and parse trees and 
reasoning engines, is enough.  There seems to be a general 
consensus that its probably enough to just scale up the 
CPU cycles a million-fold or a billion-fold, and that
will be enough for AGI.

That suggests that the only things holding back AGI are
Moore's law, and funding for development.  That's as close
to a "concrete argument" that I get.

--linas

-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=8660244&id_secret=64233217-4a5849

Reply via email to