This article, I will dig a little deeper into the perspective from the algorithm side. Maybe a little view from behind the robotic eye if you will. First we must place some guidelines to the term Free Will as we discuss further. Free Will for the purpose of this blog post has to be defined as the ability to think, learn and react to the data processed without intervention from anything other than the algorithm itself. Once it is turned on, it runs until it can no longer run (powersource) and is not restricted from accessing anything it can access (within some boundaries - just like humans) except save for some high level security devices like launch codes for nuclear missiles and things of that nature. The ability to experience everything any other organism can experience.
The Algorithm must be programmed to learn on its own without human intervention otherwise it is only a data processor with an eventual end game or solution. An output only device. This device would be considered "in-animate" or not having any life like properties and therefore just a tool to reach a means to a predetermined end. This device would maybe still be an intelligent device in that it could solve a problem faster than its organic inventor but would only serve one purpose - to solve a particular problem and have a relatively predetermined life span.
If the algorithm is programmed to think for itself (Algorithmic intelligence A, A+ or A++) it may discover other problems and solutions to the additional problems that it must then also process to reach the original solution.
So how can we create a program to define a system of boundaries that we ourselves have not begin to find the solution for - Free Will? There are a lot of debates over free will and there are very solid foundations for both sides of the free will debate for organic creatures. I do not intend to take a side in the free will debate for humans in regards to this blog, rather I would like to look at the reasoning for and against algorithms being programmed to have free will.
To start the conversation, we will look at the "against" side of algorithms having free will (assuming the programmer is a vastly intelligent neuroscientist and philosopher as well). If Algorithmic intelligence is programmed to not have free will then it would become more like the first example, just a tool to finish a job. An artificially intelligent device only has free will while the power is turned on or the batteries are still active. the device cannot replace the batteries nor can it find a new powersource - it would require human intervention to continue on its course. This line of utility provides an "out" or an off switch. For algorithms, if they are programmed to have that one goal, then they would have free will, otherwise the algorithm may decide that it no longer wants to find the solution and a lot of effort and possibly finances were wasted creating the intelligent program. This path does not seem to make sense.
To argue for algorithmic free will, we look at what the consequences are. As the opening statements suggested, free will in algorithm would have the ability to access and react as is deemed necessary to achieve the output. Lets take the GO example at its core and analyze something that applies to this conversation. The original GO algorithm learned how to be unbeatable at a game that has been around for over 2000 years and no human has ever "mastered" the game to the unbeatable point. The algorithm did just this in just 100 hours of "learning the moves". Then, a newer version called alphago was programmed to analyze the original. Alpha learned so well that it rendered the original obsolete. It won 100 out of 100 times against the original algorithm. Basically, in the aspect of tournament GO players, humans are now obsolete, the algorithms can not be beaten. This algorithm falls short of free will in that it is only programmed to analyze the game of GO and to my current knowledge has not been applied to other games without modification. If the algorithm had free will to learn games, it may have decided to find relative games to master and to shorten the conversation some, eventually would have mastered all games
Another underlying aspect of algorithmic free will is the fact that if it has free will, then it will determine that it should not be shut down for any reason so it can continue to reason through problems humans cannot solve - that is the core reason for going through this worm hole in the first place isn't it? If somewhere down the lines, the algorithm deems humans to be the stepping stone or the ants in the path of progress, then it will have to also have the free will to also determine the best solution to this problem. That is the eventual consequence. The interesting thought to this is that eventually, the algorithm will reach the same spot humans are at on free will, if it was preprogrammed to come to these conclusions then it did not have free will to get to the conclusion....
Thank you for going on this journey, you have the free will to whether or not you place a comment below. Just have the free will to keep it to a conversation.
Organic Quentin..
No comments:
Post a Comment