Microsoft Word Kurzweil, Ray The Singularity Is Near doc

We can show that there are broad classes of problems that 
cannot be solved by any Turing machine. It can also be shown that Turing machines can emulate any possible 
computer (that is, there exists a Turing machine that can solve any problem that any computer can solve), so this 
demonstrates a clear limitation on the problems that a computer can solve. Yet humans are capable of solving 
these problems, so machines will never emulate human intelligence
. Humans are no more capable of universally 
solving such "unsolvable" problems than machines. Humans can make educated guesses to solutions in certain 
instances, but machines can do the same thing and can often do so more quickly. 
•
The "criticism from failure rates": Computer systems are showing alarming rates of catastrophic failure as their 
complexity increases. Thomas Ray writes that we are "pushing the limits of what we can effectively design and 
build through conventional approaches." We have developed increasingly complex systems to manage a broad 
variety of mission-critical tasks, and failure rates in these systems are very low. However, imperfection is an 
inherent feature of any complex process, and that certainly includes human intelligence. 
•
The "criticism from 'lock-in' ": 
The pervasive and complex support systems (and the huge investments in these 
systems) required by such fields as energy and transportation are blocking innovation, so this will prevent the 
kind of rapid change envisioned for the technologies underlying the Singularity
. It is specifically information 
processes that are growing exponentially in capability and price-performance. We have already seen rapid 
paradigm shifts in every aspect of information technology, unimpeded by any lock-in phenomenon (despite large 
infrastructure investments in such areas as the Internet and telecommunications). Even the energy and 
transportation sectors will witness revolutionary changes from new nanotechnology-based innovations. 
•
The "criticism from ontology": 
John Searle describes several versions of his Chinese Room analogy. In one 
formulation a man follows a written program to answer questions in Chinese. The man appears to be answering 
questions competently in Chinese, but since he is "just mechanically following a written program, he has no real 
understanding of Chinese and no real awareness of what he is doing. The "man" in the room doesn't understand 
anything, because, after all, "he is just a computer," according to Searle. So clearly, computers cannot 

---

understand what they are doing, since they are just following rules
. Searle's Chinese Room arguments are 
fundamentally tautological, as they just assume his conclusion that computers cannot possibly have any real 
understanding. Part of the philosophical sleight of hand in Searle's simple analogies is a matter of scale. He 
purports to describe a simple system and then asks the reader to consider how such a system could possibly have 
any real understanding. But the characterization itself is misleading. To be consistent with Searle's own 
assumptions the Chinese Room system that Searle describes would have to be as complex as a human brain and 
would, therefore, have as much understanding as a human brain. The man in the analogy would be acting as the 
central-processing unit, only a small part of the system. While the man may not see it, the understanding is 
distributed across the entire pattern of the program itself and the billions of notes he would have to make to 
follow the program. Consider that I understand English, but none of my neurons do. My understanding is 
represented in vast patterns of neurotransmitter strengths, synaptic clefts, and interneuronal connections. 
•
The "criticism from the rich-poor divide":