When humans invent and create, they often rely on unexpected discoveries. Can computers help us have more luck?
An American bicycle repairman pulled a bicycle inner tube out of the tube box and handed it to his customer. The two men then started chatting, and the mechanic started playing with the inner tube box. When he grabbed the two ends of the box and twisted it back and forth, he found that it could smoothly change the shape of the box. (You can try this in a toothpaste box.) It was a small discovery, but it could change the world.
The bicycle mechanic was Wilbur Wright. At that time, he and his brother Orville Wright were developing aircraft. A big problem they encountered was how to control the flight attitude of the aircraft. The twisting inner tube packaging box inspired the Wright brothers to control the flight attitude of the aircraft by twisting the shape of the wings.
In the process of invention and creation, accidental discovery always plays an important role. In addition to the invention of the airplane by the Wright brothers, there are many such examples. For example, in 1948, Swiss engineer Georges de Mestral returned from a walk with his dog and discovered that the dog was covered in grass seeds with many small hooks. This unexpected discovery led him to eventually invent the Velcro. And in 1951, Harry Coover from the Eastman Kodak Company was looking for an insulating polymer for airplane cockpit covers when he discovered a chemical material that was so sticky that it would stick to everything. Later, he realized that this kind of thing could be used as glue, and so universal glue was born.
The stories of these inventors are indeed fascinating, but relying on accidents to create inventions means that things often go without progress for a long time. Hundreds of years have passed, but people still use very ancient methods when making inventions and creations. It is inefficient and cannot keep up with the pace of other industries.
However, now we have begun to learn to create good luck for ourselves. Those rare moments of inspiration can be gradually replaced by computer software before long. In other words, computer-aided invention is coming our way.
In the 1990s, a research team led by John Koza, a computer scientist at Stanford University in the United States, conducted groundbreaking research on computer-aided inventions. They used a computer algorithm that simulates evolution in nature, a so-called "genetic algorithm." When using genetic algorithms, the computer treats a solution as a genome, with genes within the genome representing various parameters such as voltage, focal length or material density. To start, the computer will randomly generate some genome samples from the initial gene pool, although they may not be a good solution. By crossing these genomes and introducing "variations", each offspring contains some traits from the previous generation as well as some new traits that may be beneficial. These offspring genomes are then given a given task to test them against. The best ones are selected and used as the gene pool for the next round of breeding. In this way, the process is repeated again and again, and like natural selection, the best solutions will eventually survive.
Coza's team conducted some tests to see whether the genetic algorithm could reinvent basic circuits like those developed by Bell Labs in the 1920s and 1930s. converters, amplifiers and feedback control systems. In the end, their test was successful. Their genetic algorithm could reinvent classic Bell Labs circuits.
If you think they are just lucky, there are many examples of their success. For example, they used this genetic algorithm to try various combinations of six eyepieces used in various optical instruments. Their algorithm not only reproduced all optical systems but also improved on some of the original designs. Among them, this kind of improvement can be used to apply for a patent.
Genetic algorithms are highly versatile. At the Genetic and Evolutionary Computing Conference held annually by the Association for Computing Machinery, researchers will demonstrate inventions made using genetic algorithms.
He is now planning to use IBM's supercomputer Watson to analyze millions of documents to help the software gain a deeper understanding of patents and technical papers.
Another Swiss technology company called Iprova has also invented "computer accelerated invention" technology to help inventors think laterally, and the source of ideas goes far beyond patent documents. The company won't reveal exactly how its software works, but a 2013 patent states that its software will provide customers with "suggested innovation opportunities," sources not limited to patent databases and technical journals, but also blogs. , online news sites and social networks.
What’s particularly interesting is that Iprova’s software can also change its recommendations as trending topics change on the Internet. This way, it will lead to more fruitful suggestions. Iprova's software can produce hundreds of high-quality inventions every month, and then the company will communicate with customers, who can choose some to apply for patents. Iprova's customers mainly come from the healthcare, automation and telecommunications industries. One of its customers is Philips, a multinational technology company. From this we can see that Iprova's business is booming.
Computer inventions run into patent trouble
However, if computers do most of the work, this sort of thing could undermine the patent system itself. The current patent system requires that an invention may only be patented if a "person of ordinary skill" believes that the invention is not obvious. But if the inventor only invented it using a computer, then the invention may be regarded as an obvious output of the computer and may not be eligible for patent application.
Now, this concern has become serious in drug development. If drug discovery tools become so powerful that researchers are only responsible for monitoring the software's activities, making the entire drug discovery process a no-brainer, will patenting still be possible? Therefore, computer-aided invention technology may cause many inventions to lose patent protection, which will greatly discourage inventors.
But no matter what, in the process of future inventions, computer assistance will play an increasingly important role. Computers will accelerate technological progress and level the playing field so that many people can become inventors. Although it is often said that opportunities always come to those who are prepared, but now, with the help of computers, we can greatly increase the probability of unexpected discoveries. In other words, computers can help us gain more good luck.