Re: IA vs. AI was: Longevity vs. Singlularity (fwd)

Robert J. Bradbury (
Fri, 30 Jul 1999 07:42:35 -0700 (PDT)

~> Eugene Leitl <> wrote:

> writes:
> > I agree. Why waste the benefits of diversity and *unique* informational
> > structures (us) in the sole pursuit of more raw materials for computronium?
> Because the atoms of your body could crank a lot more of diversity
> much more quickly than they do right now if rearranged
> properly. That's the rational agenda, in case the entity which devours
> you happens to be rational.
> Resistance if futile.

I agree with Eugene's atomic argument, but I would probably disagree regarding the motivations of "rational entities". This goes back to the whole question of "Why do we still see stars?". I think it revolves around the fact that there are diminishing returns on any strategy if carried out to its ultimate limits. In the example of the conversion of a solar system into computronium -- you get a *huge* jump by simply disassembling a single planet to harvest the full solar output. [~10^13 increase [earth-power output -> sun-power output]). You get ~10^8 increase in computation density by going to nanotechnology (~10^5 more IPS in ~10^3 less volume). Disassembling most of the other planets may only get you a factor of ~3-4 in terms of increased computational capacity [due to the fact that a multi-layer Matrioshka Brain has diminishing returns with additional layers]. The extra material does help in terms of long-term memory storage capacity, but it isn't clear at this point whether this is really needed [you have to make an argument that there is some "rational" purpose to remembering many details non-essential to survival].

Once you have gained something like 10^21 in processing capacity, the gains of "re-engineering humans" are going to be so far down on the scale that they will likely not merit much consideration. The *real* issues are *what* do you think about and *what* is the optimal computing architecture to use to do that. Earth is far more likely to get removed because it stands in the line-of-sight for a highly parallel commuication lasers between to computation nodes than it is because the SI needs us measly earthlings for computronium. Given the tradeoff between bouncing the beams around the Earth (relative to all the other communication delays in a MBrain) and nostalgic attachment to Earth, I would think that "saving" Earth wins.

But I can't promise that this would be the case... Much depends on how much of the MBrain is uploaded human consciousness vs. self-evolved AI.