For a long time, men want to see whether lifeless computers can exercise judgment, make choices, give birth to ideas, and play s inteltly. At the far limit of possibility, they wish to know whether, at least in theory, machines can reproduce themselves. In other words, are we really certain that a machine can do only what its programmer wishes it to do Already there are in existence a number of machines that approach these powers. Tile st type is the computer that understands logic. Samuel Butler, the great English author, predicted the view of the future world of machines in his novel Erewhon, published in 1872. Ever since, authors with less sense of satire (讽刺) than Butler have created tales of monster (怪物)machines that take over and destroy the world. In varying degrees, their imaginary monsters have the power of reproduction, which permitted them to multiply until they could capture and crush mankind. Science today is not laughing at such extremes. Dr. J. Von Neumann, of the Institute for Advanced Studies in Princeton, very carefully considered the question: Cmn machines become highly organized enough to reproduce themselves His conclusion was that they may, provided they are complicated enough. Mathematically, a certain degree of complication seems to be necessary for reproduction. Below that, the iron mothers can only give birth to more primitive forms than themselves, and any "race" of robot monsters would soon lose the reproductive function. If mathematicians are to be believed, we can rest fairly secure for a while yet. Human physiology (生理学) is still millions of times more complex than any giant brain. To copy the complexities of the human brain alone would be a shocking problem. Transistors (晶体管) might reduce the super-giant to the size of a house and his power requirements to a few hundred kilowatts -- yet the human brain will fit into a cereal bowl and takes about twenty-five watts of power. In the third paragraph, "Science today is not laughing at such extremes." suggests that______.