In by the power of super-smartness and how

In this book we can find the potential development and influence of superintelligence machines in
the future. This book focuses on a hot topic nowadays, that how should we treat with the high
developed superintelligence machines in the future, especially when they are more capable of
dealing with all kinds of intelligent tasks than human being. In the book the author specified the
advantages of human being during the evolution of species: We are not the fastest creatures on
this planet, but we invented cars, trains and airplanes. Although we are not the strongest, we
invented bulldozers. Our teeth are not the sharpest, but we can invent harder tools than the teeth
of any animal. The reason we control the earth is that our brains are more complex than the brains
of even the smartest animals. If the machine is cleverer than humans, then we will no longer be
the master of this planet. When all this happens, the machine will move beyond humanity. The
human brain has some of the functions that other animals do not have in the brain. It is these
unique features that give our race the dominance. If the brain of a machine exceeds human beings
in general intelligence, then this emerging super-intelligence can be extremely powerful and out
of control. Just as the fate of gorillas now rests more in human hands than in their own hands, the
destiny of the human future will also depend on the super-intelligent behavior of the machines.
Moreover, the author believes that super intelligence will be a huge threat to our humanity. In this
book, the author talks about the risks posed by the power of super-smartness and how mankind
can address that risk. The author believes that the issues he refers to in this book will be the
greatest risks to our humanity.
1. The potential risk that may be brought by superintelligence and how to deal with it.
In my opinion, one of the most interesting part lies in the conflict between the increasing demand
of superintelligence and the potential threat it may bring. “AI control problem” is much more
difficult than we thought, because the solution might involve instilling the superintelligence with
goals that are compatible with human survival and well-being. Solving the control problem is
surprisingly difficult because most goals, when translated into machine-implementable code, lead
to unforeseen and undesirable consequences. Hence, we can use other methods to control the
superintelligence. On the one hand, “Capability Control”, prevents the system from interacting with
the outside world freely from the physical level. Although nowadays maybe we can not imagine
how to make superintelligence come true through the whole brain simulation, and how to use
different materials to replace the biological structure and functions of our brains, we are supposed
to prevent prospective human dominance from the potential threats of superintelligence. Without
the direct interaction with the physical environment, the machines can only handle the tasks that
have already been selected by manipulators. And according to specific standard the tasks will be
filtered to decrease the potential risk. On the other hand, “Motivation Control”, The three laws of
robots. For example, some special rules must be programmed in the brain of the robots to keep
the human dominance: The robots are not allowed to hurt the human. Although there will be all
kinds of problem about how to handle the interaction between the robots and the human beings
and how to treat the superintelligence of the robots, etc. In my opinion, the prospect may not be
so permissible like the author said. The emergency of superintelligence for humans is not only
opportunities but also challenges. Society and the world will face major changes, but I believe the
ultimate trend will spiral go forward.
Furthermore, the methods of maintaining survival of robots are relatively different from the way
of humans. For instance, humans need food and water to maintain their fundamental life
movements, in contract to this the robots “eat” electricity. The fundamental requirements of
humans and robots are not conflicting, although electricity also takes a decisive part in human lives.
At last but not least, we can also restrain thoughts of robots through the limitation of the system’s
intelligence or access to information. If superintelligence gains a decisive strategic advantage, the
values we choose will determine how cosmic resources are disposed of. The risk of making wrong
choices is especially high when you are unfamiliar with the context of decision making; obviously,
the ultimate goal of machine superintelligence to shape the future of mankind as a whole is
precisely the unfamiliar decision-making background. In order to avoid hijacking the fate of
mankind, it is necessary to design a list of its options for review and approval (the main purpose is
to reduce the probability of catastrophic errors).
2. Is the prospect of achieving this type of Superintelligence realistic?
In my opinion, now it can not be specified exactly that how to realize superintelligence step by step.
I can only say that the key point of this kind of realization is the anthropomorphic thinking. When
the robots have similar thinking with the human being, they are also capable of perceiving the
surrounding environment and looking for sources to maintain their survival. In the book the author
contributes the birth of anthropomorphic thinking of the robots to the prospective whole brain
simulation. Nowadays we use a lot of machine learning algorithms based on huge amount of
training data to train the robots, in order to let they learn some new functions by themselves.,
although this thought is also too simple and not complete. And another key point lies in the
independent thoughts of robots. Using machine learning the robots can imitate the humans
movements to handle the specific tasks and actions, but they are also lack of independent thoughts
that can complete all kinds of tasks without the order of others, distinguish the advantages and
disadvantages of each task and decide its own movements by themselves. These problems are also
very difficult to solve.
Moreover, people are genetic tools. Therefore, people do not aim for themselves, but for the
purpose of preserving themselves and passing on their servants. In this sense, all kinds of human
drivers, including emotion and reason, serve this purpose. But intelligent robots do not need this
kind of driving force at all. It is just an information processing system, seeking the best solution and
exploring the ultimate truth. In other words, it will stop the stupidity of mankind and guide the
development of mankind. However, it will not rule human desires nor will it breed its own interest
unless it is based on an optimal existence plan. Even so, it is not an out of compassion for humanity
but an objective consideration for maintaining the diversity of the universe and reducing the
suffering of life. Without the desire of breeding and human emotion, robots can only act as a
computer, but never truly replace the human dominance, because they can only make decisions
according to specific rules and regulations.
In the ever-changing modern science and technology, all R & D team are all at their best, competing
to develop a new generation of robot products. If human beings can not take precautions and
scientifically respond, they will regret it later.