XplanaZine

MYTH OF SISYPHUS — Can machines replace humans, Part 2
Can machines replace humans? Most people would emphatically say no, at least not yet, covering up their disconcertion. But there are already people fearing for their jobs, fearing a machine will replace them as soon as such a machine becomes available, sooner or later. The real life examples of such replacements are not sufficient evidence to justify this fear, as we have seen in Part 1. We should have independent reasons for believing that machines could eventually do every type of work that humans can do. But the prevailing human vision about the future of machines is the opposite of the aforementioned fear and is found in almost all futuristic, science fiction movies. It is the vision that machines could eventually take on necessary tasks that people are not interested in, at least not for long periods of time, because their repetitive and tedious nature robs us of valuable time and energy we could invest in more creative and/or more pleasant pursuits. In addition, these machines would outperform humans in their assigned tasks. This vision represents the genuine human desire to live a better and fuller life.
So, the root of the anxiety should be found elsewhere. Star Trek’s Mr. Data - the likable, friendly android who performs his duties with precision and reliability - provides a great example that can illustrate some of our own attitudes. What makes Data so likable is his unassuming behavior. He is also honest, predictable and consistent and he stands by the people as a true friend by virtue of his program’s unwavering moral code. However, Data has no emotions, and yet, in many instances, his social behavior surpasses that of humans. Data behaves mostly like a human being. He may not laugh because he finds no reason for it and not because he lacks the ability to make the appropriate sound (the fact that his laughter sounds unnatural was obviously a deliberate attempt to minimize confusion about his non-humanness in the audience.)
The audience usually has two compartmentalized beliefs about Data. These beliefs don’t mix, they don’t affect one another and they don’t clash.

1. The first belief is formed by an analytic intellectual process and professes that, in spite of his unsurpassed physical strength, computational power and civil behavior, Data has no emotions simply because he is a machine, an android. Axiomatically, machines have no feelings or emotions and so neither does Data, at least not until he gets enhanced by the installation of an ‘emotions chip.’
2. The second belief is rather an attitude laying out the foundation for a pre-belief and is based on a projection. It is the same projection of our own inner states on other people who, perhaps, look like us, behave like us and thus, we conclude, they must ‘be’ like us, that is, as human as we are. We don’t know exactly what other people are feeling until we acquire some external clue, either behavioral or verbal - unless of course, we want to start including the kind of empathy that Deana Troi’s character provides. Behavioral clues are necessary but given that the same outward expression may correspond to different internal states the right interpretation becomes crucial and very difficult to determine what the deciding rules should be.

It is the two beliefs/attitudes Star Trek fans have toward Data that I believe illustrate and explain much of our current beliefs about machines and, above all, about humanity.
The creation of an android for the screen comes in distinct layers depicting the way we understand ourselves. After all, creating an android is a feeble human attempt to replicate ourselves using our understanding and personal labor. Roughly and not necessarily in this order, we think of the following steps. The first step is to make it look human, the second is to make it capable of reason, the third is to make it perform some tasks and the fourth is to make it capable of socializing by providing a program for its social conduct. The emotive part becomes harder to add to the rest. But it was tempting to imagine it as an extra programmable layer and postulate the possible existence of an emotions chip.
These steps of constructing an android may not describe our life experience as a whole, but they roughly describe the way we ‘understand’ or evaluate ourselves. That is, we perceive ourselves as bodies with reason, capable of performing tasks and of social behavior. Our own feelings should be somewhere in there but it is not clear how they fit with the rest of the picture. Are they a separate layer, or perhaps a hue, a shade, a pigment, a form? Or, could they be the brush that was used, or the canvas, or the painter’s motives, or the picture’s meaning, or the context? There are so many and radically different possible choices. If we find out how this standard picture came to prominence we can also find out whether it can include our emotive side and how.
So, where did these layers representing the way we understand ourselves come from? The key is in the second attitude mentioned above. It is precisely the result of having adopted an observer’s point of view. We believe that there is an objective truth, or, at least, that reality accepts some objective description. Therefore, objectively described facts about ‘other’ people are considered to be a more reliable source and guidance about the reality of humans. We see other people’s bodies, we have evidence about their reasoning powers and some evidence about their internal states. But these internal states are not so easily verifiable -if they can be verified at all. If Data was programmed to simulate the outward manifestation of human emotions then no one would truly know whether he actually lacked emotions, not even his creator. Our lack of knowledge about how they came to be would never prove their absence. It’s one thing to ask whether something is true and another to ask ‘why’ it is true (if the question still makes sense).
The observer’s point of view offers many advantages especially in issues of measurability but it can’t give a complete, holographic picture because the first person point of view is nowhere to be found. One of the most serious mistakes one could make is to view himself only from a distance and to evaluate his life and work only by what others can see, or, more correctly, by the way he views himself as if he was an observer. This again, affords us some consistent and measurable evaluation but it also removes from the picture the most intimate truth about ourselves - what is a both unique and necessary. The third person point of view - what was used in the android construction - has no room for subjectivity, but subjectivity is neither optional nor unreal. It is wrong to think we must choose one or the other.
The truth lies in both and is, in principle, similar to the two dimensional images we call 3D stereograms where the same drawing presents two different but interwoven pictures. One image is easy to see but the other usually needs some training and extra concentration. The 3D illusions attest to how making changes in one image necessarily changes the other, but also how neither image exists without the other.
Looking at only one of the images, viz., the objective, can explain why some people can easily imagine that a machine could do all their work. That is because it’s easy to create an image, in fact, even a random drawing counts as one of lesser value. But regardless of how intentional or complex an image may be, it doesn’t automatically acquire the second layer that our human work and beingness contain. There are only a few two-dimensional pictures that are also 3D illusions, equivalently, there are only very few things described objectively that have an inherent subjective nature - to imagine otherwise would be a delusion.
It is true that in some cases we even want to have a machine do part of our work. However, the totality of our work and existence is a dual picture that no machine can simulate because it lacks the subjective aspect that humans have. To think I could be replaced by a machine is to think that I am already a machine, albeit a sophisticated one, and this fallacious assumption prevents me from having the stereoscopic vision necessary for knowing the truth. We make computers capable of learning, we create neural networks, we can even make them say “I”, but we can’t make them have a self and the very special kind of unity it requires, even though we create components that look like they might be constituting one. And none of these processes can make up for the holographic picture the ‘I’ creates for both itself and others.