Can a Robot have Intelligence?

Many advocates in the study of artificial intelligence claim that a robot can possess genuine understanding and intelligence provided that it has an appropriate formal program and the right inputs and outputs. In this essay however, I shall argue that this claim is false on the grounds of John Searle's Chinese Room thought experiment, in which syntax is not sufficient for semantics. The robot reply to this argument, along with the claim of chauvinism made against biological naturalism only displays the irrelevancy of each argument to show that a robot can possess genuine understanding and intelligence. It is clear that no matter how the Chinese room is altered, or whether the robot is made of inorganic matter, the fact remains that a robot can only process syntactical information without understanding it and therefore does not have any intrinsic intentionality. I shall conclude that in relation to conclusions derived from the thought experiment, a robot cannot possess genuine understanding and intelligence.

According to Crane, the Chinese Room thought experiment was supposed to show that a formal computer program can "never constitute genuine understanding or thought, since all computers can do is manipulate symbols according to their form." (Crane, The Mechanical Mind, pg 125), thus refuting the stance of 'strong' AI.  The thought experiment proceeds like this: Searle is locked in a room with two slots for the input and the output. From the input slot, he receives meaningless squiggles and with the help of an English rulebook, matches them up with another batch of meaningless squiggles to feed back through the output slot. Unknown to him, he has just answered questions in Chinese, a language he has no understanding of. To native Chinese speakers outside, he appears to have understood and answered the questions appropriately. However, he had no understanding of the Chinese at all; he was just manipulating the squiggles according to the rulebook (the program) he was following in order to produce the answers. From this, it can be derived (Wilkinson, Minds and Bodies, pg 207-208) that:

Axiom 1: Computer programs are formal.

Axiom 2: Human minds have mental contents.

Axiom 3: Syntax by itself is neither constitutive of nor sufficient for semantics.

Conclusion 1: Programs are neither constitutive of nor sufficient for minds.

If we apply this argument to the case of a robot which has a formal program, it would be clear that a robot would not be able to possess genuine understanding. Like a computer, it would manipulate symbols, but would attach no meaning to them. However, as we will see in the next section, advocates of AI would say that robots would be able to possess genuine understanding and intelligence if provided with the right sensory-motor abilities, and not just a formal program with the right inputs and outputs.

The robot reply to Searle's Chinese Room thought experiment first concedes that syntax would not be sufficient for semantics, and a robot would need causal relations with its environment in order for manipulation of formal symbols to gain meaning. Its advocates argue that by placing the digital computer inside the robot's head and implementing sensory apparatus such as a camera for its eyes, a robot would thus be able to understand the syntax or match it with objects in the world and possess genuine understanding and intelligence through use of its sensory inputs and motor outputs. This argument, however, misses the point and does not address the processing of syntactical information by robots. If Searle was put inside the robot's head, and Chinese symbols were coming through the sensory apparatus, the information would still be meaningless because the robot is still manipulating formal symbols in order to give 'instructions' to the robot's arms and legs to move in response. The process of information for the process is purely syntactical, and the Chinese Room thought experiment says that syntax is not sufficient for semantics.  Thus, Searle claims, the robot is only "moving about as a result of its electrical wiring and its program" and therefore has "no intentional states of the relevant type." (Rosenthal, The Nature of Mind, pg 514) Hence, adding sensory-motor capacities to a robot would make no difference whatsoever to its capability to possess genuine understanding and intelligence.

If robots are unable to attribute semantics to syntax, what makes us humans able to do so? According to Searle, the answer is 'biological naturalism'. The human brain, he argues, "does not merely instantiate a formal program, it also causes mental events by virtue of specific neurobiological processes." (Williams, Minds and Bodies, pg 211) Because robots lack these neurobiological processes, robots are thus unable to possess genuine understanding and intelligence. This argument, however, seems inadequate and attracts accusations of chauvinism. Why is biological matter able to have the right causal powers for intentionality, and not inorganic matter like silicon? How does biological matter possess those causal powers in the first place? In her essay, Margaret Boden attempted to bring these issues to light, arguing that intentionality on the basis of neurobiological processes should really be "grounded in empirical discovery" (Heil, Philosophy of Mind, pg 257) rather than our intuitions on the subject of genuine intelligence and understanding.

In response to this claim, Boden's argument is ultimately irrelevant to whether a robot can possess genuine intelligence and understanding. While Searle's explanation for intentionality is dubious at its time of publication, the original Chinese Room thought experiment was not set out to imply that only biologically based systems could possess genuine understanding and intelligence. To clear up this misunderstanding about the thought experiment, he sets out four more axioms and conclusions (Wilkinson, Minds and Bodies, pg 211-212):

Axiom 4: Brains causes minds.

Conclusion 2: Any other system capable of causing minds would have to have causal powers equivalent to those of brains.

Conclusion 3: Any artefact that produced mental phenomena, any artificial brain, would have to be able to duplicate the specific causal powers of brains, and it could not do that just by running a formal program.

Conclusion 4: The way that human brains actually produce mental phenomena cannot be solely in virtue of running a computer program.

In sum, it can be seen that a robot with an appropriate formal program and the right inputs and outputs, as well as sensory-motor capabilities can only process syntactical information without understanding it at all. It does not matter if there are causal connections made with its environment, or whether it is made out of a certain type of material. The fact remains that without the specific causal powers for which the human brain possesses, it is impossible to attribute intentionality to the robot. Therefore, on these grounds, a robot cannot possess genuine understanding and intelligence.

Bibliography

Crane, T. (1995) The Mechanical Mind. Harmondsworth: Penguin Books

Heil, J. (2004) The Philosophy of Mind. New York: Oxford University Press

Rosenthal (ed.) (1991) The Nature of Mind. New York: Oxford University Press

Wilkinson, R. (2000) Minds and Bodies. New York: Routledge