Your ideas, money, memories, and entertainment are dreams in the minds of computers. But the thoughts of each computer are not simple, they are layered like our own minds. Their lowest, most primitive layers are the instincts of the machine. Middle layers perform more general functions of its silicon mind. Higher layers think about overall concepts. Unlike us, the computer has languages for every layer. We can teach it new ideas by changing any one or all of its layers of thought. We can tell it to consider vast and convoluted concepts. But if we make a single mistake in our instructions, the mind of our digital slave may crash in a virtual epileptic fit. When our silicon students are so pedantic, how can we engineer their thoughts to make them reliable and trustworthy assistants? And if their thoughts become more complicated than anything we can imagine, how can we guarantee they will do what we want them to? . . . Light poured in through the large windows of the lecture room. The sound of scratching pens from nearly thirty distinguished engineers and scientists accompanied every word spoken by John Mauchly. One fellow by the name of Gard from the Wright Field’s Armament Laboratory seemed to be especially diligent, writing hundreds of pages of notes. It was Monday morning, a warm mid-summer day of 1946, some three years after his stimulating tea-time discussions with Turing. Claude Shannon was three weeks into the eight-week course at the Moore School of Electrical Engineering, in the University of Pennsylvania. It had been an honour to be one of the select few invited to hear lectures on designing electronic digital computers. This was the first ever course to be taught on computer science, and Shannon was finding many of the ideas highly stimulating. He’d recently learned a new word from Mauchly: ‘program’ used as a verb. To program an electronic computer was an interesting concept. He was also hearing about some of the politics: apparently two of the lecturers, Mauchly and his colleague Eckert, had resigned from the university just four months ago because of some form of disagreement.
Research Group Introduction: Hamagami Laboratory, Dept. of Math., Phys., Electrical Engineering and Computer Science, Graduate School of Engineering Science, Yokohama National University
