Re: Economic Growth Given Machine Intelligence

From: Rafal Smigrodzki (
Date: Thu May 13 2004 - 00:16:48 MDT

mike99 wrote:

> Prof. Hanson's paper both predicts economic effects of future artificial
> intelligence as well as the current effects of computerized automation of
> jobs formerly performed by humans that we are witnessing today. In the
> Abstract to his paper, Hanson writes:
> "Eventually, computers do most jobs. At first, complementary effects
> dominate, and human wages rise with computer productivity. But eventually
> substitution can dominate, making wages fall as fast as computer prices now
> do. An intelligence population explosion makes per-intelligence consumption
> fall this fast, while economic growth rates rise by an order of magnitude or
> more. These results are robust to automating incrementally, and to
> distinguishing hardware, software, and human capital from other forms of
> capital."
> If Hanson is correct and wages fall as fast as computer prices do now, we
> are in for a bumpy ride during a historical transition period.

### You might want to consider whether the absolute levels of
consumption would follow the changes in wages.

Please consider this quotation, p.3., second paragraph from bottom: "A
Malthusian analysis of population is appropriate for this era, and
per-intelligence consumption levels fall rapidly. Per-human consumption
levels can rise rapidly, however, if humans own a fixed fraction of capital."


  (For purposes
> of this discussion, let's stipulate that computer prices fall slower than
> Moore's Law but track that Law.) Since consumption depends on income (which
> is mostly wages for most "people," or as Hanson has it, "intelligences"
> including non-human ones) then per capita consumption will drop.

### This stipulation is not a part of Hanson's model, and as noted
above, its result (human per capita consumption drop), is not predicted.


> with the "intelligence population explosion" Hanson predicts, there may be
> so many more new intelligences that aggregate consumption will increase
> enormously. Which is a good thing, since someone has to buy the output of an
> economy that is growing at least ten times faster than today's mature
> postindustrial economies (which average an annual growth rate of between 2%
> and 4%). If economies suddenly were to grow by 20% to 40% annually while
> individual consumers saw their incomes **dropping** there would be large
> potential for social unrest. Of course, if prices for most essential goods
> and services (food, clothing, medical care, education, etc.) were to drop as
> fast as or faster than wages did, the situation would not be dire. However,
> if the super-high-tech economy were also producing relatively expensive
> luxury goods (like "gotta have" electronics) that lay far beyond the price
> range affordable to most wage earners, there could still be a social
> stratification that might undermine social peace.

### Your analysis fits well into the conventional thinking that seems to
dominate this field (see Marshall Brain and his musings). However, it is
crucially dependent on conventional assumptions, such as the assumption
that human income will continue to be derived from wages. Historically,
the fraction of individual income derived directly from an investment of
labor (which in social situations frequently takes the form of wages),
has been steadily going down. While our ancestors in the EEA derived
essentially all their resources from labor (which perhaps left some mark
on our way of thinking about labor vs. capital investment), modern
Americans derive a substantial fraction of income from capital
investment. This trend is a reflection of the fact that capital (e.g. in
the form of machines, or organizational schemes)) is becoming more and
more important in production of goods and services. At the same time,
non-labor forms of capital are becoming cheaper in absolute terms (i.e.
per unit of output).

These trends will become even more pronounced as AI technology matures.
The fraction of output directly attributable to labor will trend towards
zero while absolute output is skyrocketing, the ownership of capital
sufficient for sustenance becomes ubiquitous, and therefore human
consumption will grow, in direct contradiction of the conventional wisdom.

The main objection to this prediction, raised by Hanson (with whom I
agree), is that the rise of uploads legally capable of exercising
ownership rights *and* psychological self-modification may result in a
situation where the vast majority of legally enfranchised sentients
evolved to put maximum value on self-replication, rather than personal
income growth. In that case indeed the average per human income might be
reduced - but this would be irrelevant to the issue of political
stability, as humans would be a insignificant minority, incapable of
disrupting the society.


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:46 MDT