IEEE Technology and Society Magazine - March 2022 - 89

would bring us all together? [While the] internet
didn't start the fire...it's undeniable that it has fostered
a sour and fragmented global polity-an atmosphere
of pervasive mistrust, corroding institutions
and a collective retreat into the comforting bosom
of confirmation bias. All of this has undermined our
greatest trick: doing good things together " [65], [66].
But it is not intelligence per se setting up barriers to
collectively " doing good things together, " it is the blindered
intelligence-driven, all out race to gain an advantage
and capitalize on that advantage, that enables " the
plight of things that have been created but not in a context
of sufficient care " at the parasitic expense of the
many who, under the illusion of doing more and more
for them, are bit by bit, qubit by qubit, skill by skill, job
by job, losing it for not using it. When Alexa does the
arithmetic, the child's brain does not. When GPS does
the navigating, the brain's hippocampus does not.
When the algorithm does the job, the human does not.
Epilogue
Citing Nick Bostrom's book, Superintelligence:
Paths, Dangers, Strategies, Richard Yonck fastens
on what we humans might tie superintelligent AI to
someone with psychotic obsessiveness and laser-like
attention. Countering the argument that we could just
include an OFF switch if the feeding on itself superintelligence
threatens our human future, he writes that:
Unfortunately, such responses completely misunderstand
and underestimate the nature of the threat
we are creating. ...This will be an intelligence that
is on a par with or even far exceeds our own. This
fact raises considerations and issues that need to be
addressed before we move forward along the path
to this technological singularity [2, p. 214], [67].
As a form of intelligence, the AI will possess what
is known in economics as a " utility function, " its
raison d'être. It will attempt to maximize its future
freedom of action by optimizing itself. According
to computer scientist, Stephen Omohundro, whose
beyond-the-norm concern includes the social implications
of self-improving AI, maximizing its utility
function leads to four primary drives: " efficiency,
self-preservation, resource acquisition, and creativity.
Exercised correctly, each of these, either by itself
or in concert with the other drives, can be used to
more efficiently fulfill a system's primary purpose "
[67, pp. 214-215], [68].
The four drives, Omohundro claims maximizes
a technology's utility function, are really no different
from what drives the evolution of intelligence.
They also drive inequality, winners and losers,
cumulative advantage-the Matthew effect-as
those who possess and capitalize on their advantage
take advantage [69]. In pursuit of maximizing
technology's own future freedom of action, Richard
Yonck, citing what Nick Bostrom called the
" malignant failure mode " writes that " we really
need to ensure that future AI is designed to remain
as aligned with human values as possible. The
alignment problem will require guidance from
us and it will be the guidance of a worthy intelligence,
which is where intelligence augmentation
comes in " [2, p. 240], [49], [70].
But what are these human values? Are they the values
of ransomware hackers? Who constitutes the " us "
with " worthy intelligence " who will guide the AI? Are
they the technical complexity consumed technologists?
And who will be the recipients of intelligence
augmentation? Will they be the rich getting richer
who can afford to be augmented at the inequality
exacerbating expense of the many who cannot?
If Omohundro's four primary drives, in fact, represent
the ongoing maximizing of AI's utility function,
the continuing advance of AI will enable its future
freedom of action as artificial general intelligence
(AGI) and then artificial superintelligence (ASI);
given the ongoing all-out global race for winner take
all AI spoils and threats trajectory, there's no guarantee
that AI superintelligence will not self-similarly
carry on what is going on right now, only far more
efficiently, ensuring its own self-preservation by exponentially
more intelligent acquiring of resources that,
as today, includes the, incapable of looking away,
ongoing raw material extraction of inner human
mental, physical, and social skill resources.
As for " the plight of things that have been created
but not in a context of sufficient care " in the matter
of rapidly advancing AI and its out-of-human-control
potential, James Barrat writes this:
Of the AI researchers I've spoken with whose stated
goal is to achieve AGI, all are aware of the problem
of runaway AI. But none, except Omohundro, have
spent concerted time addressing it. Some have even
gone so far as to say they don't know why they don't
think about it when they know they should. But it's
easy to see why not. The technology is fascinating.
March 2022
89
IEEE Technology and Society Magazine - March 2022

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - March 2022
