Machine Consciousness

Question 1

What is the difference between analysis and synthesis?

Answer 1

Analysis is basic science, whereas synthesis is engineering

Question 2

Why attempt to build conscious machines?

Answer 2

The key intention is to clarify, through synthesis, the notion of what it is to be conscious

Question 3

What are the advantages of focusing on consciousness rather than intelligence?

Answer 3

Performance payoffs. Better autonomy and freedom from pre-programming. An ability to represent the machine’s own role in it’s environment. Improve capacity for action based on inner contemplative activy rather than reactive action based largely on table-lookup (GOFAI - doesn’t allow for flexibility) of pre-stored contingency-action couplings (Aleksander, 2007)

Question 4

What does GOFAI stand for?

Answer 4

Good Old-Fashioned AI

Question 5

What is GOFAI?

Answer 5

The modelling or creation of intelligent machines centred on traditional computers, i.e. systems that process information, or symbols, according to explicitly coded rules. These symbols have no meaning (not grounded in real world) except as interpreted by the human programmer

Question 6

What are some generally agreed GOFAI requirements for machine consciousness?

Answer 6

The brain is a control mechanism and ensures the organism deal appropriately with its external and internal environments, but not all control systems are conscious (eg thermostats). Also the ability to model the world and self, and to integrate the two

Question 7

What are subsymbolic approaches?

Answer 7

Bottom-up, embodied, situated, behaviour-based AI. Rejected symbolic AI and focus on the basic engineering problems that would allow robots to move and survive. Higher intelligence and ‘mind’ can be created only by interacting in real time with real world, and thus that aspects of body are required. World provides constraints, information storage and feedback that make perception, learning and intelligence possible. No internal representations or models

Question 8

What are hybrid intelligent systems?

Answer 8

Combine symbolic and subsymbolic approaches

Question 9

What are some problems for GOFAI attempts to create conscious machines?

Answer 9

Why are some internal representations or models conscious and some not? How could a representation or model be an experience/conscious?

Question 10

What are some problems for subsymbolic attempts to create conscious machines?

Answer 10

Need to explain why and how conscious experience arises out of continuous interaction with the outside world. How to create/deal with conscious experiences that are not driven by continuous interaction with the outside world, eg imagining, dreaming, reasoning?

Question 11

Do conscious machines need to replicate brains?

Answer 11

A spectrum of positions on this question ranging from functionalists to physicalists who believe we need to get closer to physically replicating the human brain

Question 12

What is strong AI according to Searle 1980?

Answer 12

The appropriately programmed computer is a mind, in the sense that computers given the right programs can be literally said to understand and have other cognitive states

Question 13

What is strong AI according to Searle 1999?

Answer 13

The appropriately programmed computer with the right inputs and outputs would thereby have a mind in exaclty the same sense human beings have minds

Question 14

What is computer functionalism (computationalism)?

Answer 14

Mental states are computational states. Computational states are implementation-independent (it is software that determines the computational state, not the hardware. Since the implementation is unimportant, the only empirical data that matters is how the system functions

Question 15

What is the Turing Test?

Answer 15

A test of a machine’s ability to exhibit intelligent behaviour, equivalent to or indistinguishable from, that of an actual human

Question 16

What were two early computer programs that attempted to pass the Turing test?

Answer 16

ELIZA: an attempt to simulate a ROgerian psychotherapist (Weizenbaum 1966) and PARRY: an attempt to simulate a paranoid schizophrenic (Colby)

Question 17

What is the correct simulation according to strong and weak AI?

Answer 17

Strong AI: the correct simulation really is a mind (an appropriately programmed computer literally has mental/cognitive states). Weak AI: the correct simulation is a model of the mind (the principle value of the computer in the study of the mind is that it gives us a very powerful tool, eg to simulate mental processes)

Question 18

What is Searle’s Chinese Room?

Answer 18

A thought experiment that exploits the fact that computer programs can be ‘multiply realised’. Computer programs can be implemented ona diverse range of hardware

Question 19

What is the narrow argument of the Chinese Room?

Answer 19

If strong AI is true, then there is a program for Chinese such that if any computing system runs that program, that system thereby comes to understand Chinese. But I could run a program for Chinese without thereby coming to understand Chinese. Therefore strong AI is false

Question 20

What are 3 classes of reply/objection to the Chinese room?

Answer 20

The systems reply, the robot/brain simulator reply, and the other minds reply

Question 21

What is the systems reply?

Answer 21

Although the man doesn’t understand Chinese, there may be understanding by a larger or different entity

Question 22

What is the problem with the systems reply?

Answer 22

It isn’t a reply. It’s the thesis that Searle is supposed to be refuting - let the individual internalise all of these elements of the system, he understands nothing of the Chinese, and a fortiori neither does the system, because there isn’t anything in the system that isn’t him

Question 23

What is the robot or brain simulator reply?

Answer 23

A variation on the computer system, eg a computer in a robot body, or a system that simulates the entire brain, could understand

Question 24

What is the response to the robot or brain simulator reply?

Answer 24

Inside a room in the robot’s skull, I shuffle symbols. As long as all I have is a formal computer program, I have no way of attaching any meaning to any of the symbols, and the fact that the robot is engaged in causal interaction with the outside world won’t help me

Question 25

What is the other minds reply?

Answer 25

Either the man does understand Chinese, or the scenario is impossible. We attribute consciousness to humans on the basis of behaviour, so why not the behaviour of the man in the Chinese room or the whole system?

Question 26

What are the Lessons learnt from the Chinese Room?

Answer 26

Searles official argument against strong AI fails, but he does have a point, that merely implementing a program is arguably insufficient for the system having a mind. Something else is needed, perhaps certain kinds of causal connections between the system and its environment

Question 27

What two things does Dehaene et al 2017 discuss?

Answer 27

Global availability (C1) and self-monitoring (C2)

Question 28

What is global availability?

Answer 28

Information that’s selected for further processing, eg we can recall it, act upon it and speak about it. Similar to Block’s access consciousness

Question 29

What is self-monitoring?

Answer 29

Ability to monitor own processing and obtain information about oneself (introspection or meta-cognition)

Question 30

What is Dehaene et al’s argument?

Answer 30

We contend that a machine endowed with C1 and C2 would behave as though it were conscious (though Carter et al 2018 as ‘what about phenomenal consciousness?’)