Page 174 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

164 CHAPTER 8 The New AI: Basic Concepts, and Urgent Risks

Directorate after that. I accepted the neural network job at NSF in 1988, mainly
because of the arrow on the lower right here: the hope that better mathematical
understanding of intelligence in general would help us to understand not only the
brain but also the mind, on the deepest possible level, and help expand our ability
to develop the deepest human potential more effectively. At the recent conference
of the AsiaePacific neural network Society (APNNS), I responded to the last ques-
tion of the entire conference by giving my personal view: “I personally agree that the
human mind is more than just the brain of a mouse, but any true Daoist master would
understand that we cannot truly, fully understand the consciousness and soul of a
human before can understand the brain intelligence of the mouse. It is a crucial
step along the path.” Section 4 will also say more about these issues.

1.4 RISKS OF HUMAN EXTINCTION, NEED FOR NEW PARADIGM FOR
INTERNET OF THINGS
But at the end of the day, the end of the day might be coming. More and more, we can
actually understand why humans tend to underestimate the possibilities of massive
changes for the better or for the worse. There is a growing community of people
who believe that there is a serious possibility that the human species might well
deploy new technologies in a way which results in total extinction of the species,
maybe sooner and maybe later, and that we really should be working more effectively
to understand what the threats are and what could be done to reduce them.
Misuse of artificial intelligence, as in the Terminator movies, is one of those
threats. Yet as the Internet of Things (IOT) grows very quickly to control every
vehicle (civilian or military) on earth, every generator, every home, every city and
every medical device, in a coordinated way, perhaps from physically hardened phys-
ical servers, the threat of artificial stupidity (AS) is almost as large, and far more
imminent, than the threat of AI. A new relation is being built very quickly between
humans and the web of computer software which will be controlling them and their
economy. It is a matter of life or death that this new relation not start off on a bad foot,
fatal to all sides. That requires development of a new paradigm for the IOT which
includes not only a focus on safer forms of deep learning, but also better interfaces
and coordination with unbreakable operating systems, market-like distributed con-
trol, teleautonomy, advanced quantum technology, and, above all, new energy to
make sure that deep human collaboration and deeper human potential are enhanced
and empowered in ways that allow humans to keep up with it all. Section 5 will give
my current groping thoughts about these life or death challenges (see Fig. 8.2).

2. BRIEF HISTORY AND FOUNDATIONS OF THE DEEP
LEARNING REVOLUTION
2.1 OVERVIEW OF THE CURRENT LANDSCAPE
All of the big firms in Information technology (IT) now have substantial activities in
“the new AI” based on deep learning and neural networks. The New York Times has