From: Billy Brown (bbrown@conemsco.com)
Date: Tue Apr 27 1999 - 10:14:48 MDT
GBurch1@aol.com wrote:
> Speaking of future timelines and scenario building, we had a discussion
here
> about a year ago on general scenario-building that resulted in my
compilation
> of a timeline stretching from 2000 to 2015. You can find it at:
>
> http://users.aol.com/gburch2/scnrio1.html
>
> I have rechecked it from time to time and, while it is of course still
quite
> early, I'm still pretty happy with what we came up with.
I notice that your predictions of computer speed and storage density are
noticeably below what an extrapolation of current trends would predict (i.e.
you have a doubling time of ~2 years, rather than the observed 1 - 1.5). Is
this deliberate?
I also see a few points to quibble with:
> circa 2005
> 50 Gbt hard drives common, being replaced with completely solid state
devices
Solid state devices are improving, but not any faster than hard drives. I
don't see desktop machines switching to solid-state unless hard drives stop
improving so fast, and that doesn't seem likely to happen so soon. People
will only trade their 50 GB disk for a 1 GB persistent RAM device in a few
niche markets, like PDAs and wearables.
> coding of most applications automated
If you mean using code generators to do some of the grunt work, sure (I'm
doing some work on that myself). But figuring out how to write a novel
program is a process of invention that can't be reduced to algorithms of any
reasonable complexity. A general-purpose code writing program is thus only
slightly simpler than a sentient AI, which makes this an extremely
optimistic projection. I'd guess more like 2015-2020.
> first cell-phone implants
Wouldn't this have to go through FDA product-safety testing? If so, add
about 10 years.
> return of "oil shocks" as third world hydrocarbon consumption skyrockets
Our current oil reserves are so large that even very strong consumption
growth won't send prices up all that much. Finding and exploiting new
fields is a much cheaper, faster process than it used to be, and we can't
possibly exhaust our total supply in this time frame. At worst, we might
face shortages several decades in the future.
> a few "intelligent freeways" begin operation in major urban areas
God, I hope not. Can you imagine what will happen the first time there is a
system crash? Instead of a small, human-error accident every few days, you
get a huge, photogenic, machine-caused disaster every few years. Even if
the overall mortality rate is lower, the public won't go for it.
> circa 2005 - 2015
> The AI Revolution arrives: Toward the end of this period autonomous agents
> become important economic actors, treated as human agents are now in law
and
> business
If you mean sentient AI this seems a little optimistic (I'd bet on
2015-2025), but not hopelessly so. However, IMO non-sentient AI will be
both more common and more important than human-level sentient agents. It is
also easier to do - it is far simpler to duplicate a single cognitive
ability than an entire mind.
Billy Brown, MCSE+I
bbrown@conemsco.com
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:03:38 MST