In an issue of the French magazine La Nature dated June 7, 1890, I found an article about a competition for the design of a tower for the Paris World Exhibition in 1889.
In an issue of the French magazine La Nature dated June 7, 1890, I found an article about a competition for the design of a tower for the Paris World Exhibition in 1889. The winner, of course, was Alexandre Gustave Eiffel, for his eponymous structure. No less than 68 designs were submitted; 24 of these were illustrated in the magazine article. The designs all bore some similarity to the Eiffel Tower but were different nonetheless, since every designer has a unique and recognizable design style.
This leads me to the subject of creativity. Creativity can lead to a diversity of solutions for even a narrowly defined problem.
From the point of view of thermodynamics, when a person creates something that wasn't there before, we call such a situation "noise"; there is an "output signal" that we cannot infer from the input signal.
Of course, this is a playful comparison that does not do justice to creative people, but it means that if we want computers to be creative, then they must have a method that enables them to be creative. They need a smart program for smart, nontrivial solutions. And for that to happen, people must first understand the "how" of their own creativity.
I find M.C. Escher's iconic "Drawing Hands" lithograph, depicting two hands that appear to be illustrating each other, a splendid representation of the idea.
One of the first "computers" was a weaving machine from the late 18th century. It could weave an image that was previously programmed by punched holes in a card; with that card, several woven images could be made. Later, this punched-card idea was adapted by IBM founder Herman Hollerith to devise a machine that could count, for use in the 1890 U.S. census. Hollerith's innovation slashed the time required to process the census data from years to months.
The machine's capability was later expanded to include the ability to add. In 1928, IBM brought the Type IV Tabulator--which had even more capabilities--to market. The rest is history.
I discussed this topic in a lecture at the University of Groningen (Netherlands) in 1997 and also in an earlier course concerning this subject. The reasoning went as follows: The brain has some 100 billion neurons. That leads to a number of possible interconnection patterns. Assuming each neuron can fire 10 to 20 times a second, I estimate the brain's information processing capacity to be on the order of 1023 bits per second.
If we want to reach this speed with a chip, then the question (aside from the one that asks whether we will ever truly understand the nature of creativity) is: When can we produce a chip with that processing capacity?
For my estimate, I applied Moore's Law (assuming that somehow we will be able to continue scaling technology according to its precepts) and assumed that the progress of computer architectures would result in a continued increase in the processing capacity by a factor of 10 every five years or so.
Next, we consider how many five-year periods it would take to reach the required processing speed. The first machine with 1 giga-operations per second appeared in the market in 1997. Therefore, we cannot expect to have the required capacity before 2070 or so; in other words, it would take 14 five-year periods. With special computer architecture, we might have it a few years earlier.
The current top-of-the-range computers include the IBM Blue Gene/L, with a processing speed of 360 tera- flops or 360,000 Gops (vs. a 1-Gops machine in 1997).
The creative computer may not become a reality anytime soon. But the principles of creative behavior and learning are already being applied in clever ways in our browsers, our communications, our planning tools, our user interfaces and . . . well, use your creativity! This is an area in which entrepreneurs can find new applications and innovative products. It all depends on our human creativity and perseverance.
Being creative requires us to be pragmatic at the same time: While we strive for perfection, let's not try to reach it immediately.
Cees Jan Koomen, formerly an executive with Royal Philips, is an entrepreneur who is involved with several startups, venture capital and building an early-stage ecosystem in the Netherlands.