|Intelligent Systems And Their Societies||Walter Fritz|
There has been quite a bit of discussion between two schools of thought; those stating that humans have free will and those saying that there is no free will, that all is predetermined. This has special importance for the question of whether humans are responsible for their actions.
Here we will review the mechanisms involved in artificial IS's. In biological IS's the functions are equivalent but the mechanisms are different, because neural fields perform them. There is a choice. But past experience limits this choice; in other words, the artificial IS can only choose from those response rules, that exist within its memory. But the choice is more limited than that. Because, as we have seen in Selection of Responses (For continuous reading, like a book - do not enter here now)., of all existing response rules, the IS puts only those into the "short list", that have (in their situation part) some element in common with the present situation. (The "short list" is a first selection or pre selection). For instance it will not include a response rule pertaining to electricity when looking for the right movement for swimming. No elements of the present swimming situation exist in those response rules on electricity.
Once the artificial IS has established the short list, it evaluates all included response rules: How much of the situation part of a response rule exists in the present situation? Are there elements in the present situation not covered by the response rule? Does the response rule have elements not existing in the present situation? What was the past success of the response rule? And so on. Then it makes a choice, at random, but weighted by this evaluation. It chooses the more likely response rule more often and the less applicable response rule seldom. So a choice exists, but it is a weighted, random choice and past experience limits it. If this is what you mean by "free will", then the artificial IS has "free will".
What is the corresponding function in the biological IS? Again there are response rules, encoded in the neural fields. Also here the choice is by weighted chance; the same person, in the same situation, does not always choose the same response. Sometimes he or she tries something else.
The concept of "responsibility" does not seem to be too applicable to this process, but a penalty, very definitely, is of use. A penalty changes the probability that the IS selects a certain response rule, by making some response rules unsuccessful and thus making their future selection less probable.
So there is a kind of "free will", a choice, but the IS makes the choice by weighted chance, based on experience. Punishment and reward influence this choice by making the results of the response rule good or bad for the IS. The above insights into this problem should have effects on making and applying laws.
For continuous reading, like a book - continue
Jump to the e-book Contents / Ethics as a Science document / top of this page.
Copyright © New Horizons Press