Souls

One familiar sf scenario is a decades-long interstellar voyage with passengers in suspended animation tended either by robots or by a rotating crew. In one of the chronologically confusing chapter-long flashbacks of Use of Weapons (London, 2013), Iain M Banks relocates his protagonist, Zakalwe, as a passenger and temporary crew member on such a voyage. This is another interesting change of scene but I am having trouble pulling the whole narrative together.

Another crew member tells Zakalwe a story. Because his people discussed the question of the soul so much, a philosopher king ordered a global debate after which he withdrew to think, then published two books, one endlessly repeating, "Souls do exist. Souls do not exist," the other endlessly repeating, "Souls do not exist. Souls do exist." (p. 356)

This might be considered either amusing or profound but I saw it as both wasteful and insulting to religious believers, philosophers and psychologists. Banks could better have spent those two pages summarizing the history of the subject, starting with people inferring a soul from the experience of apparently leaving the body while dreaming. I commend the Buddhist anatta teaching but am interested to hear reasoned expressions of other theories or beliefs.

Winning A War

In Use Of Weapons, it is a thing of beauty when Zakalwe takes charge of a war run by incompetents and turns it around. A province with a dozen important religious sites is about to fall. Defending the province will only get the sites damaged or destroyed whereas surrendering the province now means that they will be regained intact with much new treasure when the war is won.

Surrendering the province will also stretch the enemy's supply lines while the rains start behind them. Enemy commanders will know that the surrendered province is a trap but will be ordered in anyway, which will demoralize them. Also, by stopping the bombardment of roads, Zakalwe ensures that the roads will fill with refugees, slowing enemy troops further. But bombing some bridges and the enemies' oil refineries will make their commanders advance more cautiously even if, as believed, they have great reserves of fuel. In any case, Zakalwe distrusts his side's intelligence about enemy fuel and thinks that the enemy are equally uninformed.

When the enemy Imperial Court and their high command meet, Zakalwe proposes dropping an otherwise defunct and useless spacecraft on them, pretending that it is the first in a series of new missiles, then offering to negotiate with the commoner's parliament while the enemy are riven by civil war.

One man turns everything round - but then is told that larger scale politics require his side to lose.

The Artificial Intelligence Question In Use Of Weapons

An artificial intelligence addresses a human being:

"'Your brain is made up of matter...organized into information-handling, processing and storage units...'"
- Iain M Banks, Use Of Weapons (London, 2013), p. 279.

The AI lists three factors that have organized the matter of the brain in this way:

genetic inheritance;
biochemistry;
experiences since before birth.

The AI asks:

"'An electronic computer is also made up of matter, but organized differently; what is there so magical about the huge, slow cells of the animal brain that they can claim themselves to be conscious...?'" (ibid.)

Nothing magical but we are in fact conscious. Otherwise, we would be unable to "claim" anything. Also, the term, "'...experiences...,'" entails consciousness.

"'...but would deny a more finely-grained device of equivalent power...a similar distinction?'" (ibid.)

Many devices made of organized matter are not conscious. However, if "'...of equivalent power...'" means "...that performs exactly the same functions...," then, yes, this electronic computer is conscious.

There is a verbal ambiguity here because previously the term "electronic computer" applied to artifacts that merely manipulated symbols without any knowledge of their meanings and therefore were not conscious. Thus, a biological organ that merely handled, processed and stored "information" could conceivably be an unconscious organic computer rather than a conscious brain. The information stored in a book or library is not conscious. An intelligent conscious being is any entity, organic or artificial, that is capable not merely of scanning and copying texts but of reading and understanding them. "Understanding," like "experience," entails consciousness.

Before asking its question, the AI had said:

"'Forget...about how machine brains are actually put together...'" (p. 278)

But how they are put together matters. Surely that determines whether or not they can be conscious? The phrase "'...machine brains...'" is ambiguous. If it means "machines that duplicate the functions of brains," then these machines are conscious whereas if it means "machines that simulate the functions of brains," then such machines are not conscious.

The AI continues:

"'...think about making a machine brain - an electronic computer - in the image of a human one.'" (ibid.)

Any artifact that is exactly modeled on a human brain so that it perfectly reproduces the functioning of such a brain will, by definition, be conscious in the way that a human brain is but will not be what used to be called "an electronic computer."

The AI suggests this process:

"'...start with a few cells, as the human embryo does'" (ibid.);
let these multiply and connect;
add new components;
make connections "'...identical...'" (ibid.) with those of the stages of human development;
in order exactly to duplicate human development, limit the speed of transmissions along the connections;
have "'...these neuron-like components...'" (ibid.) fire messages in response to signals received;
thus, exactly mimic the development of a human brain ("mimic" is ambiguous between simulation and duplication);
also mimic its output by sending signals similar to sound, touch and light as experienced inside the womb (surely this is further input, not "output"?);
simulate birth;
apply sensory stimulation so that the device thinks that it is "'...feeling, touching, tasting, smelling, hearing, and seeing everything your real human was...'" (p. 279).

At the end of this process, the AI suggests that the device can be given "'...just as much genuine sensory input, and of the same quality, as the human personality was experiencing at any given point.'" (ibid.)

Does "...the human personality...at any given point..." mean a particular person whose experiences will be directly transmitted into the device? I do not find this last point entirely clear but, in any case, if the device is genuinely an artificial brain that is literally experiencing sights, sounds etc, then of course it is conscious. But that is because the AI has described an exact duplication of human development. This still leaves open the question of the nature of those machine brains whose construction we were told to forget about in the first place.

Although the AI refers at one point to "outputs," its description of the developing device is of an entity that is almost entirely passive in its reception of inputs. In this respect, it appears to duplicate the development of a human embryo and infant. However, I think that organisms became conscious not merely by receiving inputs but primarily by interacting with their environments. Thus:

organisms were naturally selected for sensitivity to environmental alterations;
organismic sensitivity quantitatively increased until it was qualitatively transformed into conscious sensation;
sensation was naturally selected because pleasure and pain have survival value.

By "sensation," I mean the most elementary stage of consciousness, the transition from being hot to feeling hot. This happens because an organism approaches life-giving heat and avoids dangerous heat so that action, not mere reception, is paramount. Since "sensation" entails consciousness, the phrase "conscious sensation" is redundant but I use it to differentiate sensation from sensitivity that has not yet become conscious. A sensitive recording device is not conscious but sensitive organisms became conscious.

The AI asks:

"'...where is the difference? The brain of each being works in exactly the same way as the other; they will respond to stimuli with a greater correspondence than one finds even between monozygotic twins; but how can one still choose to call one a conscious entity, and the other merely a machine?'" (ibid.)

What is the point of spelling out that two entities are identical and then asking what is the difference between them? The question still remains how are AI's in the Culture constructed and are they conscious? Of course, Banks writes their dialogues in such a way that they clearly pass the Turing test but these books are works of fiction. We have not yet encountered such entities in reality.

What does "'...merely a machine...'" mean? So far, machines have been mechanisms designed to perform functions without being conscious of them. In a motor vehicle, consciousness is provided by the driver, not by the engine, and, when a computer merely manipulates symbols, then consciousness is provided by programmers and users, not by the computer. If, however, some other "machines" can duplicate human consciousness, then they are conscious machines, not "mere" (unconscious) machines.