HL extension: Philosophy and contemporary issues (Paper 3) practice questions.

Turing set aside the question whether machines can think and proposed instead a procedure in which a judge, reading typed answers alone, tries to tell a machine from a person. The philosophical import of this replacement is that

• an observable criterion is put in place of a claim about inner states
• thought is shown to require a brain made of biological material
• the question is settled by inspecting the code from which the program was built
• the speed of a system's replies becomes the mark of understanding

Answer: an observable criterion is put in place of a claim about inner states

Turing offers a criterion of performance, not an account of consciousness. Searle's Chinese room is precisely the objection that the behaviour tested for could be produced by a system that understands nothing.

A critic replies to Searle that a program housed in a robot with cameras and moving limbs, causally hooked up to its surroundings, would understand what it says. Searle's answer to this robot reply is that

• a body would help only if it were built from biological material like ours
• the added perceptual input reaches the program as further symbols to manipulate
• the robot understands, even though the person inside the room does not
• a system with sensors and limbs could no longer be simulated by a person in a room

Answer: the added perceptual input reaches the program as further symbols to manipulate

Searle asks us to imagine him inside the robot, receiving the camera's output as further uninterpreted shapes and sending back shapes that move the limbs, so causal contact with the world adds inputs rather than meaning. Defenders of the reply answer that grounding in the world is precisely what mere symbols lacked, which is why the exchange turns on what meaning requires.

In debates about the moral status of artificial systems, Bentham's question about animals, whether they can suffer, is invoked to argue that the ground of moral status is

• membership of the human species, which artificial systems cannot acquire
• the ability to pass a test set by human judges
• the capacity to suffer rather than the capacity to reason
• the possession of legal rights conferred by a competent authority

Answer: the capacity to suffer rather than the capacity to reason

Bentham's question makes moral patiency turn on sentience, so a system could be highly capable and still lack moral status, while a non-intelligent sentient animal has it. Whether a being that can be set back without being able to suffer also qualifies is a further and disputed question.

A commuter keeps every appointment in a phone, consults it automatically and relies on it without checking. A defender of the extended mind thesis would say that

• the phone is a tool that the commuter's memory reaches for, not part of it
• the case shows that human memory has been weakened by the device
• the phone counts as part of a mind only if it is conscious of its contents
• the stored entries already function as standing beliefs of the commuter

Answer: the stored entries already function as standing beliefs of the commuter

Clark and Chalmers argue that a resource that is reliably available, automatically endorsed and easily accessed plays the functional role of memory, so its entries count as the agent's standing beliefs, whatever the hardware and without being conscious. Treating the phone as a tool that memory merely reaches for states the rival embedded view, on which external resources support cognition without belonging to it.

A delivery system that keeps training on its own operating data injures a pedestrian, and a court orders the operator to pay damages under a strict liability rule. Someone who holds that there is nevertheless a responsibility gap here is committed to saying that

• the payment allocates a cost without showing that anyone is blameworthy
• the ruling was mistaken and the operator should not have been made to pay
• the system itself should have been named as the defendant in the proceedings
• the gap closes as soon as an insurer covers the pedestrian's losses in full

Answer: the payment allocates a cost without showing that anyone is blameworthy

The claim concerns the conditions under which a person may be blamed, namely control over the outcome and knowledge of what one is doing, and not the question who can be made to bear a cost. Liability is allocated by rule and can always be assigned, so a decided case leaves the moral question where it was.

A recruitment model is trained on a firm's past hiring decisions and reproduces the pattern of those decisions with high accuracy. A philosopher replies that this accuracy settles nothing about fairness. The best reconstruction of that reply is that

• the model's predictions are not accurate enough to be relied upon
• statistical claims cannot appear in a moral argument at all
• an accurate description of past practice does not justify continuing it
• fairness requires that every group of applicants be hired in exactly equal proportions

Answer: an accurate description of past practice does not justify continuing it

The move from how things have been done to how they ought to be done needs a further evaluative premise, which is the is-ought gap. Equal outcomes for every group is a substantive theory of fairness, not what the objection asserts.