Can machines think?
Cogito, ergo sum! We, as human beings, intuitively ensure ourselves that we are capable of thinking, so our existence. Nonetheless, for the last decades, we have had this controversial argument about whether computers can think. The argument is still progressively being raised as artificial intelligence and sophisticated computers instill in our lives more and more [1, 2]. People from several disciplines, such as philosophy, computer science, and mathematics, reflect on this conundrum; while some of them are in favor of one side of the argument, some others acknowledge that whether computers are qualified to think is a mystery. Turkish mathematician Cahit Arf passed an opinion about the way our brain works in his representation ‘Can a machine think and how can it think?’ in a public conference on artificial intelligence in 1958 [3]. According to the representation, the brain works as a chain process. We can briefly conclude that the brain stores and then analyses external data by using logical reasoning, inferences with analogies, several calculations, and finally, a relevant outcome exported through our distributor organ, also the information about the process is stored, if necessary. After introducing how the brain works, he stated that machines can think, the machines are deprived of aesthetics compared to humans, though. We now have more insights about computers and we are closer to the assumption that computers, i.e. machines, think. For me, a sound approach to progress over an argument that I am being challenged with is to start from opposing ideas so that I could reflect on antitheses. The main issue that computers are frequently questioned about and stated as insufficient regarding human-level thinking is creativity. I will consider creativity from two different viewpoints, the first one is about the creativity patterns, whereas the second viewpoint addresses a comparison between human-like intellect and computer capabilities.
Can we consider inspired works that endow and enrich accomplished ones that are being inspired as creativity? Thanks to Generative Adversarial Networks (GAN), an artificial neural network developed by Ian Goodfellow and his colleagues, creativity is all around us [4]. We can design a poster for a seasonlong series of 9 million video frames [5], create new works by interpreting a set of foundations based on the paintings created by the artists, and paint human faces that do not exist in reality [6]. The Creative Adversarial Network (CAN) takes GAN to the next level [7]. It achieves more than enough creativity and credibility to make computers construct art that is indistinguishable from the art of humans. Besides artistic creativity, AlphaGo blends creativity and analytical skills and beats the world champions [8]. AlphaGo was developed by the Google Deepmind team and it has the potential to convince most opposing views on creativity. When Lee Sedol commented on AlphaGo’s famous 37th move “I thought AlphaGo was based on probability calculation and that it was merely a machine. But when I saw this move, I changed my mind. Surely, AlphaGo is creative!” [9]. So what is this 37th move? AlphaGo made a playable move with a probability of 1 in 10.000, which seemed like an unreasonable move at first to experts and spectators, even making the software team think there was a problem [10]. Later on, it appeared as a coherent and sensible move that would not even occur to players with 100 years of experience. It’s proved that AlphaGo did not just replicate previous games and did not proceed with advanced computations, but it also had an intuitive aspect that could not be fully explained similar to the phenomenon of competence, creativity, or mind. The team’s latest work is MuZero’s is being developed a solution-oriented structure with decision-making and without domain or game knowledge and human data [11]. These developments delicately delineate how computers are close to learning how to learn.
Further, it would not be fair to expect human-like creativity from computers, as we do not know the exact functioning and psychological foundations of it on human-level. Some researchers even argue that there is no such thing as creativity [12]. This view prompts us to reflect on the differences between computers and human beings in terms of mind-related skills, creativity, intelligence, and intuition rather than if computers are capable of these skills. We know that there are capabilities in which humans perform better than computers and vice versa. Jolly et al. [13] conduct a study about the intuition skill of humans and the lack of this skill on computers. They consider intuitive physics that concerns the ability that humans and, to a certain extent, infants and animals have to predict outcomes of physical interactions involving visible to the naked eye objects. Such differences may primarily arise because human beings almost always interact with real life and all the things around them through their sense organs [14]. To illustrate, we can reflect on our sight. We may not realize the point our vision has reached thanks to evolution, almost half of our cerebral cortex is reserved for the visual functioning and has a complicated structure [15]. While visual reasoning is a basic operation in our daily life, they are quite challenging for computers. To exemplify, while the sequence of photos depicting the stages of the dishes we see in recipes is apparent to us, it can turn into a difficult task for computers [16]. The technical leader of IBM Watson, an artificial intelligence tool, Rob High endorses computers’ incapabilities and these differences between computers and humans in a useful way. He considers computer intelligence as a revelation for human creativity. He appraises and appreciates computer intelligence by saying “It is not our goal to recreate the human mind that’s not what we’re trying to do. What we’re more interested in are the techniques of interacting with humans that inspire creativity in humans, and that requires that we spend time thinking about that creative process”.
As a final thought, the argument of Can computers think? emphasizes the scope of abstract concepts, creativity, intelligence, and intuition, along with the accomplishments of the computers so far. Computers may not think like humans for now, but this does not assure us that they do not think, and should not contempt the analytical issues in which they are far superior to humans. We should particularly examine the reasons behind computers’ inability to think like human beings. One of the main reasons for this shortcoming of computers is that computers have much less and limited data than humans obtain as they cannot keep continuous interaction with the real-world [14]. It wouldn’t be wrong to relate computers to a small baby in terms of possessing and interpreting data, but we cannot ignore the thinking ability of computers, just as we cannot ignore the baby’s thinking ability. In addition, while the functioning of the concepts of intuition, creativity, and thinking mechanism in the human brain is still ambiguous [12], it would be flawed to measure how well computers perform in the simulation of these concepts. In conclusion, we should deal with this much deep and abstract issue better if we focus on a relatively narrower and well-defined scope, still, our frame at hand implies that there is a great chance for computers to be able to think provided far more real-world interaction than they have as of today.
References
[1] Emma S Brunette, Rory C Flemmer e Claire L Flemmer. “A review of artificial intelligence”. Em: 2009 4th International Conference on Autonomous Robots and Agents. Ieee. 2009, pp. 385–392.
[2] Achim Hoffmann. “Can machines think? An old question reformulated”. Em: Minds and Machines 20.2 (2010), pp. 203–212.
[3] Cahit Arf. “Can a Machine Think and How Can It Think?” Em: Atatürk University — Disseminating University Studies and Public Education Publications Conference Series. 1958–1959, pp. 91–103. url: https://www.mbkaya.com/hukuk/cahit-arf-makine-dusunebilir-mi-orjinal.pdf.
[4] Ian J. Goodfellow et al. Generative Adversarial Networks. 2014. doi: 10.48550/ARXIV.1406.2661. url: https://arxiv.org/abs/1406.2661.
[5] Burak Karabulut. “In The Context Of Artificial Intelligence Creativity and The Future Of Visual Design”. Em: 20.79 (2021), pp. 1–23. issn: 1304–0278. url: https://dergipark.org.tr/tr/download/article-file/1460732.
[6] Tero Karras, Samuli Laine e Timo Aila. A Style-Based Generator Architecture for Generative Adversarial Networks. 2018. doi: 10.48550/ARXIV.1812.04948. url: https://arxiv.org/abs/1812.04948.
[7] Ahmed Elgammal et al. CAN: Creative Adversarial Networks Generating “Art” by Learning About Styles and Deviating from Style Norms. Jun. de 2017. url: https://arxiv.org/pdf/1706.07068.pdf.
[8] AlphaGo. url: https://www.deepmind.com/research/highlighted-research/alphago.
[9] Andre Ye. How DeepMind’s AlphaGo Became the World’s Top Go Player. Mar. de 2020. url: https://ai.plainenglish.io/how-deepminds-alphago-became-the-world-s-top-go-player-5b275e553d6a.
[10] The challenge match. url: https://www.deepmind.com/research/highlighted-research/alphago/the-challenge-match.
[11] Muzero: Mastering go, chess, Shogi and atari without rules. Dez. de 2020. url: https ://www.deepmind.com/blog/muzero-mastering-go-chess-shogi-and-atari-without-rules.
[12] Marvin L. Minsky. “Why People Think Computers Can’t”. Em: AI Magazine 3.4 (dez. de 1982), p. 3. doi: 10.1609/aimag.v3i4.376. url: https://ojs.aaai.org/index.php/aimagazine/article/view/376.
[13] Mary Jolly. The Concept of Intuition in Artificial Intelligence. 2014. url: https://www.academia.edu/6441833/The_Concept_of_Intuition_in_Artificial_Intelligence.
[14] Brenden M Lake et al. “Building machines that learn and think like people”. Em: Behavioral and brain sciences 40 (2017).
[15] Charles Stangor e Jennifer Walinga. “Chapter 5. Sensing and Perceiving”. Em: BCcampus, 2014.
[16] Semih Yagcioglu et al. RecipeQA: A Challenge Dataset for Multimodal Comprehension of Cooking Recipes. 2018. doi: 10.48550/ARXIV.1809.00812. url: https://arxiv.org/abs/1809.00812.
