Can machines be conscious if they are not alive?

For centuries philosophers and later psychologists have debated what consciousness is, without reaching a consensus. With the progress made in neural networks, deep learning and artificial intelligence, the debate is now reaching technologists and artificial intelligence experts. Consciousness has been debated for centuries, though there is no unique definition.

While we humans make a claim to be conscious, when consciousness arise in a child is still debated.
Similarly, while most would agree that a plant growing towards the sun does not do so in a conscious way, it would be harder to reach a consensus about the level of consciousness that more advanced animals, say cats and dogs, reach.
While defining consciousness seem to escape us, it may be easier to define what behaviours we would clearly define as unconscious and what behaviours could be clearly defined as conscious.

Any mechanical response, for example a primitive reflex, can be classified as clearly an unconscious process, so much so, in fact, that it is not possible to control. On the other hand artistic creation, mathematics, abstraction in general, seem to be mental processes that would require some kind of consciousness. However, with the progress of neural networks, many examples are being created of advanced machines that can create music or images by simply being trained on existing ones.

Magenta is an open source research project that uses TensorFlow to generate new content manipulating images and music. Yet, this kind of artistic creation would not, by many at least, be understood as an example of consciousness.

What, then, would classify as a clear conscious process? The difficulty stems from the fact that consciousness tends to be understood as something that should allow to classify each clearly defined action in time as either the product of consciousness or not. For example, when a plant moves towards the light, is that a conscious process? When a person trying to lose weight, after some thought, decides to refuse a piece of dessert, is that a conscious process? At first sight one could claim that the plant is not conscious, but the person is. However, one could also argue that the person’s action is simply the result of a choice that is derived from some internal processes that took some time to balance all alternatives and make a choice. Desires are typically intended as unconscious or the result of some conditioning. The desire for sweets has a basis in our internal brain chemistry, similarly the desire to appear beautiful (by being slimmer) is the result of some external conditioning, and therefore the choice of whether to eat or not to eat a dessert may simply be the result of complex, but still mechanical, processes that, therefore, cannot be called consciousness.

In fact, we often tend to classify as conscious most actions that seem to require some complexity, but complexity does not have to necessarily equate consciousness. An artistic process seem to be a conscious and complex process, yet we have machines that are able to create “art”.

Observing “actions” restrained in time may not allow us to have enough elements to reach a consensus on the consciousness of the being making the action. Observing, on the other hand, humans on a scale of millennia or longer, may allow us to better identify what is that we perceive as consciousness. Comparing humans to plants or animals over the same period may allow to distinguish what makes consciousness.

In this regard, humans, unlike many other living beings, seem to to have evolved, though it took millennia, the ability of creating machines that may surpass their own abilities. Computers make calculations faster than humans, cars run faster, planes can fly. In order to make something better than itself one needs to have some level of consciousness of its own self. Similarly, animals that are able to use some basic tools to improve their own abilities may demonstrate some level of primitive consciousness. However, these actions do not translate in an improvement of the species, but only of the individual’s activity, and could just be the result of the interaction of the different external sensorial stimuli with experience to produce a solution. It would suggest some creativity and abstraction, possibly, but not necessarily full consciousness.

One of the hottest debates in recent years is about when machines will be able to surpass human intelligence and, at the same time, create even more intelligent machines that will keep creating better ones. Will machines be able to do it without acquiring consciousness? The problem is that machines may be able to learn to create better machines than themselves, but that creation would simply be an “action” constrained in time, not the result of an evolutionary process. It would be the simple application of rules learned and improved. If, therefore, consciousness is a quality that can only be appreciated not on single individuals performing precise actions, but rather on all the individuals of a species that aim at improving itself, consciousness can only exist as a biological construct, not a technological one, and therefore there can be no consciousness in beings that do not experience life as a process of growing and reproducing, but they can only be complex entities performing complex actions.

Consciousness may not be, as often thought, an individual quality, but it may rather be the by-product of the appearing of life in the universe, and it may never be obtained by machines.

Leave a Reply

Your email address will not be published. Required fields are marked *