Can computers think?
Published:
In the everyday life of us human beings, we do one of the most complex processes we have ever attempted to comprehend: ‘thinking’. In a tiny amount of time, with almost no effort and almost no failure, the task is done with the help of our fundamental tool: the brain. It goes without saying that recreating human brain is a challenging task. It is not only a difficult task to make computers to think but we also don’t really know how we do it. However, the studies and research projects to reconstruct human brain should be taken into consideration as well. Moreover, the results of those projects mostly indicate that there is hope. Therefore, the question “Can computers think?” will be discussed through those studies.
The question “Can computers think?” is such a popular one in modern philosophy and it is asked by Alan Turing for the first time in his paper published in 1950. Turing is the most influential person who has reflected on this question, undoubtedly. In his paper, “Computer Machinery and Intelligence” [1], he describes the problem in terms of a game which is known as the “imitation game”, or “Turing Test” (TT) where a test person asks questions to both another human being and a computer via computer screen. If the test person can’t spot the difference between the two answers, the computer passes the test and it is said that it is able to think, in other words, display intelligence. Since the test offers a very practical solution, it is discussed by a wide majority. Some stated that:
“Turing’s paper represents the beginning of Artificial Intelligence (AI) and its ultimate goal as TT. On the other hand, some stated that TT is nonsense, even useless” [2].
In a speech [3] by Noam Chomsky, American cognitive scientist, he calls TT as “a meaningless test”. He also states that:
“to ask whether machines can think is … like asking do submarines swim?”
which was first demonstrated by Edsger Dijkstra, ACM Turing awarded scientist, in 1984 [4].
Before going any further, what is intelligence anyway or what does it mean to think? To even define what intelligence is or what thinking really means is challenging already. In general discourse, thinking means the action of using one’s mind to produce thoughts. But, TT comes up with a practical solution and what it does is to “equate intelligent behavior with human behavior. This is not necessarily the case.” [5]. Quite often the human can display unintelligent behavior such as; thinking that hitting a broken computer would make it work again.
As Isaac Asimov stated in his paper [6], intelligence is a subjective matter and it depends on the particular field concerned. Therefore, each one of us may not be expert in many fields, which is also supported by the famous quote:
“Everyone is a genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is stupid.”
So perhaps, we are testing the intelligence of a computer in a wrong way. Maybe limiting computers to means of communication such as language is the same as judging a fish by its ability to climb a tree. In addition, can we really say that human intelligence or the human way of thinking is the only way of thinking? [5]. Maybe computers have a different concept of thinking which humans cannot understand and they perform at their best in that way. Therefore, they would fail TT while passing all the intelligence tests.
Another drawback of TT is what is known as the “Chinese Room” which was first published in a paper in 1980 by American philosopher John Searle [7]. In the experiment, Searle is alone in a room with a computer and some Chinese characters are passed over to him down through the door. Even though he doesn’t speak Chinese, he manages to come up with the right strings of Chinese characters since the computer changes the symbols and numbers, which makes those outside the room believe that the person in the room can express himself in Chinese. What can be elicited from this test is that: “programming a digital computer may make it appear to understand language but does not produce real understanding. Hence the “Turing Test” is inadequate”[8]. TT may have failed but the studies on reconstructing the human brain keep gaining momentum.
The renowned physicist Dr. Richard Feynman once said:
“What I cannot create, I do not understand. Know how to solve every problem that has been solved.”
That is what we have been trying to do even though we don’t understand how the brain works. “Because the key to understanding how intelligence works is to recreate it inside a computer”, says Shelly Fan, a neuroscientist at the University of California, San Francisco. There is only one problem here: our computers cannot handle the operations done by over 100 billion interconnected neurons and trillions of synapses. But back in time, those numbers were even smaller. In 2012, researchers at the University of Waterloo, Canada, unveiled Spaun [9], a 2.5-million-neuron model of the brain that bridged the gap between neural activity and biological function. In 2013, the Human Brain Project came out which was co-funded by the European Union. Another study at the Advanced Institute for Computational Science in Kobe, Japan the supercomputer, also known as the K computer, is capable of at most 10% of neurons and their synapses in the cortex. Thanks to the Moore’s Law we know that computers will be faster and faster each year.
Previously, in February 2018, an international team re-created the structure of a simulation algorithm and made a big step forward creating the technology to achieve simulations of brain-scale networks on future supercomputers of the exascale-class. With the brand new algorithm [10], larger parts of the brain can be represented while speeding up brain simulations on existing supercomputers. “Since 2014, our software can simulate about 1% of the neurons in the human brain with all their connections” says Markus Diesmann, Director at the Jülich Institute of Neuroscience and Medicine (INM-6). “In order to achieve that success, the software requires the entire main memory of petascale supercomputers, such as the K computer in Kobe and JUQUEEN in Jülich”, says the author at Frontier in Neuroinformatics where the work was published. “With the new technology we can exploit the increased parallelism of modern microprocessors a lot better than previously, which will become even more important in exascale computers” adds Jakob Jordan, lead author of the study, from Forschungszentrum Jülich.
Moreover, in May 2018 Google announced its brand-new technology behind a new Google Assistant feature: Google Duplex [11], a new technology for making general conversations to carry out real-world tasks over the phone such as scheduling certain types of appointments as shown in the paper by the voice messages.
In the lights of all the aforementioned innovations, improvements, it is obvious that we have come a long way. Even though we have a lot more to go, we proceed day by day to our goal of recreating the human brain. I can clearly say that I am excited about the future and I believe that the human brain will be created one way or another.