Re: AI Researchers: Are They Evil? (fwd)

Robert J. Bradbury (bradbury@www.aeiveos.com)
Wed, 4 Aug 1999 00:49:13 -0700 (PDT)

> paul@i2.to wrote:

Minor request...

Unless I go back and read the early messages in this thread, I can never remember what IA's or CRNS stand for... If you happen to be a fast typist and the thread run time is long could you from time to time expand the acronyms?

> Eli, you want to create an SI before nanotech destroys
> everything.
This unlikey to happen. Nanotech creates the SI. We can create an SI without Nanotech (10^24 Merced+'s sucking up the Sun's power output), but it requires either (a) A government program to build self-replicating non-nanoscale manufacturing plants or (b) a very long time because it is going to take most of the Si in Venus to build the computronium. Planetary dismantlement goes *slow* without nanotech.

> People like Den Otter, Max More and myself want
> to IA ourselves to singularity before the SI's destroy us!
An interesting question would be -- Since if SI's exist they certainly could have destroyed us already, they either don't exist or have an interest in not destroying us. If they have an interest in not destroying us, then it may be because they have a desire in seeing us "grow up" as a "good" SI. The *question* then becomes is a "good" SI one that embraces its former "humanity" and uploads everyone or one that leaves behind the passions of lesser creatures to become one of pure logic or thought?

Which ever case is true, since the SIs have the power to enforce their "perogative", it doesn't matter *squat* which of those alternatives you prefer.

> What makes you think you can close the gap between
> assembler and SI any sooner than we can close the gap
> between SI and uploading?
This I can answer. The assembler is the hard step. You go from nothing to SI level (10^25x a brain) in a few weeks to a few months. With that much simulation capacity, I have to believe uploads will be figured out very quickly. There would have to be a *huge* investment into creating the hardware to do uploading from our handicapped brains into suboptimal hardware and perhaps a "go slow" mandate on building the SI to get uploading before we get an SI.

Now, it just may be the case that if the chemists get self-assembling molecular electronic computers to work *before* a true diamondoid nanoassembler gets built that the situation will be reversed. In that case, you can build hardware enough to upload, without the ability to easily turn a the solar system into an SI.

In summary:
Molec. Electronic Computing before Diamondoid Nanoassembler:

==> Uploading before SI
Diamondoid Nanoassembler before Molec. Electronic Computing:

==> SI before uloading

It is interesting, that the development of these paths could easily be different from civilization to civilization and that could have very interesting consequences for SI personalities and perspectives.

Robert