Re: What is Intelligence?

Dan Clemmensen (
Sun, 29 Sep 1996 19:52:59 -0400

Robin Hanson wrote:
> I think I've been sloppy in my posts on this topic (especially the post that
> Peter McCluskey responded to), so let me try to do better.
Yor posts are not usually sloppy, at least to me.

> The three obvious contributors to the intelligence of the world economy are
> population size, the raw speed of human brains, and knowledge (both tacit and
> explicit). While there are other constributors, such as some high level
> organizational issues, I doubt they are that important, and expect progress
> on improving them to be slow.

I feel strongly that speed of access to information and speed of
of new information are very important contributore to the equation.
> Knowledge and population have been increasing for some time. As the world
> economy grows, the more we know, the better we get at learning more, and the
> larger population we can support. And being richer, we've been able to spend
> more on research, and on tools to aide research. (Tools are really
> artificats that embody knowledge, in my view.)
> Computers are just one of those tools, and haven't really aided research much
> more than other important tools, like cars and air conditioning. Progress in
> computers is driven by the size of the market for computers - the more
> computers folks buy, the more researchers are employed, and so the more ideas
> get tried out.

Computers are a new phenomon. Their impact is only now being seriously
So far, the contribution to knowledge has been mostly in the
Communications, word processing, publishing, etc. More recently,
lab insturments have dramatically increased progress in some fields,
genetic sequencing. E-mail and the web are now the communications media
choice in several disciplines, including branches of math, astronomy,
physics. but also in other areas. Comuputers are now essential elements
in the
development of new digital circuits, including new computers. Thus, I
that the situatin you describe is changing very rapidly.

> Thus we have had a feedback process of growth of knowledge (and tools) and
> population for some time. The big change I see looming here is that of
> decreases in the cost, and increases in speed, of individual brains with
> uploads. And especially the ability to make copies, which decreases the cost
> of filling new brains with knowledge (the big limit on our growth rate now).
> Since this should drive down the cost of labor, research gets cheaper, and so
> more should be done.

I prefer not ot predict the exact nature of the way computers will
intelligence, since IMO it's not central to our discussion. I feel that
uploading is considerably harder than other approaches to human/computer
collaborative entities.
> As our technology of uploads improves, for the same price we can either make
> faster brains or more brains. It is not obvious which option will be more
> preferred - the most important thing isn't brain speed but brain cost per
> speed. In particular, research may get done by a few very fast very
> expensive brains, but I'd guess lots of slower brains to be the usual case.
If uploads are similar to other types of algorithms, faster will be
to more, because multiple processors spend a lot of time on
of intermediate results.

> Anyway, I basically agree that a change we can envision will increase the
> growth rate, but I think folks are confused about the root cause. The basic
> issue isn't brain speed, nor cost per speed. Human brains are actually
> pretty cheap out there at the moment. The big change is lowering the cost of
> creating knowledge-filled brains, with upload copies.
I feel that "knowledge-filled" will be superceeded by the ability to
the entire knowledge base at a very high data rate.

> If average brain speed goes up as the growth rate, the subjective effect to
> the participants may not be that different from today.

This is likely to be correct. The "Singularity' affects only the meat
humans, not
the digital ones, assuming that the extended interacting algorithms
remain in the
form of individual uploaded humans. my guess would be that this is not
the most
efficient or the most likely form of the algorithm, however.