IEEE Technology and Society Magazine - September 2019 - 31

But the process doesn't stop with us. It carries on in the
accelerating concentration of power as escalating exergy
(see footnote 1) - "negative entropy" as the Nobel laureate physicist, Erwin Schrödinger, coined it [44] - in
advancing technology, with its supreme manifestation in
cutting-edge artificial intelligence.
Two recent articles in IEEE Technology and Society
Magazine, "Efficiency Versus Creativity as Organizing
Principles of Socio-Technical Systems: Why Do We Build
(Intelligent) Systems?" by Ada Diaconescu [38] and
"Assessing Artificial Intelligence for Humanity" by
Andrzej Nowak, Paul Lukowicz, and Pawel Horodecki
(NLH) [39], fasten on the growing concern that the feeding-on-itself advance of AI and its cutting-edge sidekicks
(machine learning, deep learning, neural networks,
robotics...), pose an increasing threat to our human
future unless corralled into serving humanity and not
on a plate.
Both Diaconescu's and NLH's essays zero in on the
nature of technology as a double-edged sword.Our
human future keys on whether AI, with its emphasis
on efficiency above all else with adverse human consequences as the unfortunate externality - or IA,
"intelligence amplification," NLH's "Human Centric
AI", dominates in our future. En route, they make recommendations aimed at realizing the latter. But what
is the likelihood of the realization of AI enhancing
rather than extinguishing us? In their essays, the
authors keenly capture the dark side of the sword in
many of its facets: the valuing of efficiency as an end
in itself rather than a means to a better human end;
and the by-and-large unrecognized slippery slope
towards the parasitic sapping of human mental, physical, and social order due to technology increasingly
eliminating the need to use brain, body, and communication skills. Towards remedy, Diaconescu recommends paying keen attention when the language of
efficiency is being used to justify whatever is being
promoted, marketed, and sold as a not to be questioned value in itself regardless of suitability, targets,
beneficiaries, case-by case. Efficiency for what? Efficiency for whom? "[What] are the potential side
effects?" "Are the benefits worth the potential risks?"
"[What] does an 'efficient education', 'efficient
research?', or 'efficient healthcare' even mean?"
"[When] patients must be released as soon as possible lest they become negative statistics for a hospital's performance evaluation, what kind of efficiency
is being optimized, and for whom - certainly not for
the patients; nor for the community." Pay attention,
she warns, to the kinds of technological devices we
develop and/or use. Ask not only what it can do for
us, but also what, along the way, is it doing to us?Use
technology "selectively, for specific well-suited tasks
SEPTEMBER 2019

∕

rather than by default for everything... [Do] we have to
use a smart phone [cum Alexa] to control the lights?
Are we giving away valuable data or knowledge?
Who benefits? What are the impacts?" In two words,
pay attention.
In what they call a "Function-Oriented AI" future,
where humanity increasingly looks to AI to do the
work, NLH zero in on the losing-it-for-not-using-it equivalent of waste heat in the downslide of human brains,
bodies, and societal bonds. They write that "Any skill
that is not used decays. As an increasing range of
tasks is delegated to AI, humans will lose the knowledge and skills to perform these tasks and will
become helpless without AI... In this scenario, human
information processing is delegated to AI, and
humans just get answers. They don't gain understanding of the knowledge and processing rules that led to
the solutions... In contrast, the concept of Human
Centric AI," (IA)envisions future AI technology that will
synergistically work with humans for the benefit of
humans and human society: Towards that goal they
stress enhancement not replacement, empowering
people instead of prescriptive systems telling them
what to do, transparent, validated, and trustworthy,
not "black box" systems, with human values, ethics,
and privacy embedded in AI system and app core
design considerations.
While Diaconescu's recommendations are crucial
to any chance we have of steering the juggernaut
towards uplifting human ends, the likelihood of their
real world, sustainable-on-a-species-scale implementation are slim to nonexistent for all but a sliver of
humanity on the top end of the benefit-to-harm scale.
The same goes for NLH's Human Centric AI realization
recommendations. The reason is that technology is
carrying on the self-similar path that led matter to selforganize into life in the first place. Technology is
replacing humans as a still more efficient means for
the Second law to pump out entropy, to degrade any
and all gradients, the differences that can make a difference, not excluding our individual and collective
human differences that can make an augmenting
human difference.
The reason that either article's necessary, but not
sufficient recommendations for tilting the scales
towards IA rather than its current, prescriptive, justgive-us-the-answers AI, has to do with the technologycorrupted urge to take the supremely powerful, once
vital-for-survival draw of whatever path to a goal minimizes food energy consuming exertion - "The Principle of Least Effort" [40]. The need to pay attention to
the qualifiers of efficiency - for whom, for what, with
what side effects - demand more effort than just poke
and go. This knee-jerk urge to take least effort paths to

IEEE TECHNOLOGY AND SOCIETY MAGAZINE

31
IEEE Technology and Society Magazine - September 2019

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - September 2019
