John Searles Case against Artificial Intelligence: Only a Machine Could Think - Essay Example

?John Searle’s Case against Artificial Intelligence: ‘Only a Machine Could Think’ Introduction Searle’s (1980) reply to the query “Could a machine think?” by saying, “… only a machine could think” (p. 417, 431) could be easily misconstrued to mean that Searle’s reply is affirmative. Yet, further examination of his reply would show that he is in fact strongly saying otherwise – a stance that has triggered varied counter reactions until now. Though Searle’s replies may not be strong enough to have sufficiently pinned down the persisting claim of functionalism and computationalism that machines/computers can fully think, his argument nonetheless has continually put Artificial Intelligence (AI) under close scrutiny until today. Summary of Searle’s Claim Searle’s (1980) reply to the query: “Could a machine think?” is built on two propositions, which he stated as follows: “(1) Intentionality in human beings (and animals) is a product of causal features of the brain. (2) Instantiating a computer program is never by itself a sufficient condition of intentionality.” Searle explains that the strict logical consequence of his first two propositions is (3) the explanation that the manner by which the brain produces intentionality invalidates the claim that intentionality is also produced by instantiating a computer program. Hence, a computer program cannot produce intentionality. He furthers that the trivial consequence of his first proposition is (4) the need to possess causal powers similar with those of the brain to enable any mechanism to produce intentionality. Hence, a machine should have a human-like brain to be able to think. Lastly, he explains that what follows to his propositions 2 and 4 is the proposition (5) that any literal attempt to create intentionality through artificial way would fail to do so if it will only design programs like the AI; what it needs to succeed is to recreate the human brain’s causal powers. As I understand, the first proposition which he assumes to be an “empirical fact about the causal relations between mental processes and brains,” (p. 417) implies two more specific propositions, as follows: (1a) Human beings and even animals have intentionality, and what made them have intentionality is because (1b) they have the causal powers of the brain. Hence proposition (1) can be stated as human beings have intentionality because they have the causal powers of the brain. This way of saying it is to state Searle’s argument in another way – that intentionality is the mental activity that human beings are capable of doing because of the causal powers of the human brain that they biologically possess; for a machine to think it must have intentionality which can only be possible through having the causal powers of the brain. Hence, not unless the machine has the causal powers of the brain similar to human beings, the machine could not think. Thus, to say that AI, as what functionalism and computationalism persistently assert, can fully think – with the understanding that thinking here has intentionality – is short of saying that AI can also be human beings – a claim that obviously Searle does not want to accept, rejecting every possibility that AI could think and consistently defending his position that intentionality is a mental characteristic inherent to human beings. Actually, AI’s claim seems harmless, but perhaps Searle has perceived its dangerous implication that’s why he obstinately opposes it. Hence on his part, Searle simplifies his proposition (1) in a way that does not allow an AI to become capable of thinking: “… certain brain processes are sufficient for intentionality” (p. 417). Searle’s way of simplifying his proposition (1) is to emphasize his point that intentionality requires necessary causal features of the brain that is far more than the information processing system that computationalism is so proud of or the formal symbol manipulation of functionalism, because these causal features of the brain are in fact chemical-biological phenomena that can only be replicated biologically, meaning through procreation, and not mechanically, meaning through programming. So, why did Searle argue that “… only a machine could think” (p. 417, 431), when all his arguments show that he vehemently rejects the idea that machines can think. To be clarified it would be best to take Searle’s full reply to the question regarding the ability of the machine to think. Searle replies: My own view is that only a machine could think, and indeed only very special kinds of machines, namely brains and machines that had the same causal powers as brains. And that is the main reason strong AI has had little to tell us about thinking, since it has nothing to tell us about machines. By its own definition, it is about programs, and programs are not machines. (p. 431) From his reply it can be understood that only a very special machine could think. That very special machine is the brain because it has the causal powers needed for intentionality. Hence, it is only the brain that could think, unless of course there were other machines with the same causal powers as the brain. It is on the basis of being a machine that AI could not think for the very obvious reason that AI is not a machine; it is a program; and obviously a program is not a machine. Perhaps Searle’s argument can be best understood by differentiating machines from programs. The Strengths and Weaknesses of Searle’s Position I find Searle’s propositions regarding the causal power of the brain to produce intentionality most telling in two ways. First, his argument immediately put in place the order of things – that human beings are unique irreplaceable creations and machines are meant only to serve and not to replace human beings. Although it could be argued that AI does not intend to replace human beings, the manner by which AIs are being developed and defended logically threaten human existence. Second, his argument also emphasize that intentionality is a mental state which is a chemical-biological phenomena innate in human beings, hence can only be replicated biologically and never mechanically. This further strengthens the uniqueness and irreplaceability of human beings, thus, putting in check any attempt or ambition of doing otherwise. To make his point, Searle makes use of a Chinese Room Argument. Here, Searle illustrates that computationalism is very wrong in thinking that the human mind is simply a computational device and thus the ability of the machine to compute could be perceived as thinking; and that functionalism is very wrong in thinking that by enabling the machine to follow instruction is enough manifestation of thinking. Essentially he disputed these claims by illustrating that essential to thinking is the ability to understand the meaning of symbols. Without this there can be no intentionality. Searle’s Chinese Room Argument is actually effective in the sense that it compelled his opponents to recognize the necessity of understanding in thinking. Although they appear to oppose Searle, their arguments essentially explain that contrary to Searle’s observation the machine actually understands what it does or what it is made to do. For example, the Systems Reply states: “… understanding is not being ascribed to the mere individual; rather it is being ascribed to this whole system of which he is a part” (Searle, 1980, p. 422). As Searle says, “the systems reply simply begs the question by insisting without argument that the system must understand Chinese” (p. 423). Thus, Searle’s Chinese Room Argument enabled him to bring the debate into proving that the ability of the machine to compute and to follow instruction must possess understanding in order to qualify these as thinking. Therefore, understanding is a crucial factor to thinking. In the same way, understanding is crucial in the dialogical dimension of cognition that is characteristic of human intentionality. However, Searle’s Chinese Room Argument was unable to illustrate that the machine is incapable of understanding. His argument would have been stronger if he created a scenario wherein the computer is made to understand and then failed to do so. The best way to refute the view that the machine can think is to show that the machine cannot think. Furthermore, Searle correctly identified the vitality of semantics (meaning) in understanding syntax (symbols). However, I did not find his argument to clearly illustrate what is in semantics that cannot be transported or imitated or programmed in the machine. It could have been clearer if Searle simply yet explicitly stated that ascribing meanings to symbols is not a simple mechanical process, as seem to imply by the Robot Reply, which argues that by enhancing the perceptual (putting a camera) and motor capacities (putting robotic arms) of the machine will enable it to understand – and the Brain Simulator Reply, which argues that by basing the system rules on how the human brain operates would surely enable the machine to understand. AI proponents miss the point that meanings are not mechanically but are socially constructed. When I say socially constructed, meanings are shaped by various social realities such as cultural, political, economic, or technological and hence, they are experience-based. Moreover, their construction does not occur at one time but a series of time. It is actually these changing social realities that add or modify the meanings society and individuals ascribed to symbols. I don’t think this dynamic process of human thinking could be imbedded or programmed in AI. Actually, the manner by which Searle rejects the claim of AI for intentionality is grounded on the biological make-up of human beings, which, following his assertions, cannot be replicated in robots or computers or any form of AI. As Searle (1980) states: “Whatever else intentionality is, it is a biological phenomenon, and it is as likely to be as causally dependent on the specific biochemistry of its origins as lactation, photosynthesis, or any other biological phenomena” (p. 431). This I believe is a strong argument, in two ways. First, it emphasizes that intentionality is biological, meaning it is a natural human make-up. Hence, thinking is a process natural to human beings. As such, it cannot be simulated or reproduced mechanically. Immediately this sharply defines the boundary between the human being and the machine – a boundary that AI persistently blurs. Thus, Searle is correct to say that whatever enhancements given to AI, still it could not think. Because merely by its name artificial intelligence, AI has admitted that it cannot have what constitutes intentionality. Second, this argument emphasizes the limitations that natural law imposes – something that AI advocates would like to surpass. In fact, it is along this argument that Searle rendered AI immaterial against the Many Mansions Reply, which essentially argues that Searle’s arguments applies only to analogue and digital computers which unfortunately is the state of technology during that time, but when the time comes that those processes that Searle states to be essential to intentionality are imbedded in AI, then all his arguments would be irrelevant. This reply only affirms Searle’s contention on what produces intentionality. Consequently, this has also pinned down the AI’s thesis that “mental processes are computational processes over formally defined elements” (Searle, 1980, p. 427). Though, his argument is already strong, it could have been clearer if this argument related intentionality to personhood, because intentionality in human beings defines personhood. Thus the saying goes, ‘what you think is what you are’. That is why semantics are crucial. As can be observed, this point of personhood is what is being consistently missed in all those arguments against Searle’s propositions and argument. To further argue, this I think is the weakness of the AI thesis: It fails to understand what is essential in intentionality and it fails to understand that what is essential to intentionality cannot be simulated or replicated in machines, because these essentials are natural biological processes not mechanical ones. So if AI wants to pursue its claim that it can think, then as the Many Mansions Reply has argued, it should possess all the processes that constitutes intentionality. Conclusion Although Searle’s replies may not be that sharp to pin down the counter arguments of those who think that AI could think, still, his propositions and replies have essentially pushed AI proponents backward. Though they may pursue their claim that AI could think by further developing the AI, imbedding in it those causal powers of the brain that Searle successfully presented to be necessary to produce intentionality, the matter is in fact mute and academic. Because as Searle has illustrated intentionality is a natural product of the human brain – a very special machine that is chemically-biologically constituted. It is not artificial. Hence, only the brain could think. Reference Searle, J. (1980). Minds, brains, and programs. Behavioral and Brain Sciences, 3(3): 417-457. Read More