Frankenstein: the Modern Prometheus (Book)
By ai-depot | June 30, 2002
Awaking to Senses
There’s a short section in the book where the creature describes the period just after it has received life. This is an interesting setting, since only immature brains are put through this stage in real-life (i.e. small babies). The creature has a fully developed brain and can experience this consciously, which prompts its description.
To put things into perspective, however, this is only a postulation. The author is rather more interested in the philosophical aspect of the story, which pertains more to fiction than science. Little biological research has been done on this, since very little is possible before we understand the human brain better.
The creature describes a period of a few days, through which everything appears mixed up. Dark and cold, loudness, warm but bright… All these stimuli he has now learnt to identify, and can relate them. At the time though, all this was a great mess.
After about three days, he was able to make head and tail of all these sensations. From then on, he was able to identify and perceive them, using coherently in his interaction with the rest of the world. The interesting thing to note here is that everything seemed to fall into place automatically. His senses just came into focus like when you put a pair of glasses on. This leads to believe that the brain is capable of self-organising to adapt to these raw inputs. Additionally, this is an unconscious process, as he learnt to perceive information without really applying himself to that end.
So far, in full virtual agents which have been created by researchers, their inputs consist of prepared data. A somewhat convenient representation for the artificial intelligence algorithms is used. This is part of the process which I term ‘designing intelligence’, without which the agent simply wouldn’t perform to expectations. So in essence, the low level stimuli from the environment are abstracted out from beneath the agent, allowing it to deal with higher level information. In that respect, a self-organising approach based on Shelley’s description would allow the agent to learn to deal with such stimuli, and avoid the programmer the hassle of simplifying the input for the agent.
This could be an important break-through for Artificial Intelligence. Once again, I’m using the conditional to indicate this is just based on Shelley’s extrapolation, and could turn out to be a complete mistake. However, this seems plausible, and more to the point, it seems like an approach which has potential.
Tags: none
Category: review |
