This is a summary of Chapter 1 of "Make and Move ALife". If you have any questions or concerns, please let us know in the comments.
Alife does not try to understand "what is life" by observing living organisms that already exist. We make a lot of hypotheses that "what is the essence of life?", Create an artificial system based on it, and take an approach to understand the essence of life while making it.
By linking technology and engineering with life as in genetic engineering, our perception of life is gradually changing. Life is not something that cannot be changed, but something that can be changed according to purpose. For example, if the artificial limbs (= part of life) of a physically handicapped person become more sophisticated (= changeable), it will be possible to run faster (= purpose) than a healthy person.
There is also the view that "does life have a purpose in the first place?" From the standpoint of Darwin's theory of natural selection, the process of natural selection is not purposeful. Let me give you an example. When we see a flying bird, we think that the feathers are adapted to fly. However, in reality, it is believed that the original function of the wings was not to fly, but to keep them warm. It is hypothesized that the feathers that evolved to keep warm have evolved into flat feathers that can draw various patterns by sexual selection, and that they happen to be used as wings for flying.
Alan Turing created the concept of "computable things". Based on this concept, the basic principle of a machine (= computer) that can actually execute calculations was considered. Then, is the "consciousness" we have computable? Does "artificial intelligence", which has been attracting attention in recent years, have consciousness?
To find out, let's briefly sort out how humans perceive it. The "consciousness" that we have is scrutinizing the information that comes in from the sensors (five senses) and connecting it to exercise for the body. When I look outside, it's raining (= scrutinizing the information coming from my five senses), so let's leave the house with an umbrella today (= exercise for the body). This is called "sensory-motor coupling". Deep learning is good at extracting necessary information by categorizing the information received from the sensor. But the real perception process is more complicated.
Even if you are given the same information, how you behave depends on the situation. Quoting Yukkusukuru's theory of umwelts, sea anemones for hermit crabs are food when they are hungry. If you're not hungry, it's a shell decoration. In other words, even for the same input, different outputs can occur depending on the context.
As you can see, it is very difficult to create a complicated processing part that is contextual. You need an unbuilt system that is as soft as the consciousness of life.
――Artificial intelligence and ALife (artificial life) are different. ――ALife will understand life and the umwelt consisting of that life while inheriting the concepts of "calculation", "life", and "system".
Recommended Posts