Is Government by Artificial Intelligence the Way Forward?

Are we to be governed by artificial intelligence?

Could this be worse then present political bigotry?

At present, most democratic nations are governed through some form of political party system, with the result that an election can mean a change of direction, for the economy and governing attitude towards everything. Then 5 years later it gets reversed and all political directions are changed again.

This is clearly upsets the stabilization of every aspect of life in that nation. Will we be better served by replacing political parties and the bureaucracies that actually govern us with artificial intelligence?

Computers that learn to think for themselves are already being used in insurance and banking systems, often taking away the human input to decision making. This means common sense is no longer applied, a wealthy person who has never previously borrowed and has considerable assets will be judged as a high risk if they wish to raise a loan, while a person with a history of debt will not be high risk, unless they have defaulted on a loan. This due to the computer saying no credit history equals high risk, without any other consideration.

This is the center point of considering if artificial intelligence would govern better. It is only as good as the parameters set by humans when initiating the program. The old adage of a computer saying a stopped clock is more accurate than one 5 seconds fast, still remains. The computer says the stopped clock is exactly right twice a day while the clock that is fast, is never right.

Artificial intelligence can gather data very fast and utilize it in making decisions but the data is accepted as truthful and accurate, while humans should know this is not the case. There exists already in some systems a "runaway feedback loop” and this can cause false decisions. One example is in assessing household insurance risks. The data included the post (Zip) code of the applicant and the computer is constantly acquiring data about accidents, crime, and police activity in any post code. It then can say the risks are too high and refuse cover. This leads to uninsured homes. The computer then labels the area as high in non insurance and so makes the risk level even higher as it concludes this is a deprived area and so will have a high crime rate. If the area is redeveloped and new upmarket houses and shops built, they will have difficulty getting insurance as the system will class it as high risk while any human would understand things had changed. The initial parameters should have radical change inputs programmed in but it is doubtful if this will be done.

The accuracy of date being absorbed by these “intelligent” computers is the biggest drawback since the computers can link with each other any false data gets repeated and used, as if accurate, in many different situations. Take an example of a social worker inputting data and they misspell a name, for example putting in Ross when should be Rose, the computers will collect all data about the Ross family and add the new input from social welfare and then this file of information becomes a true picture of the Ross family and this is shared with other computers for insurance, police, banking etc. and yet the data includes information intended for Rose family. The data is now false for both Ross and Rose families but the interlinked computer system will assert that it is accurate at every instant it is used in making a decision. This could have far reaching and very damaging effects on both families for generations.

The present UK system of elected representatives deciding who will govern and the USA version where representatives are elected separately from a President; can both elect flawed people. These elections often result in the personal beliefs of those elected, becoming far more involved in the governance than the long term overall welfare of the nation as a whole. Handing over control to intelligent computers, who theoretically have no personal bias and could be programmed to deal with everything logically and for the greater good in long term; sounds appealing but this is also a very high risk structure of government. Who sets the initial program? What data is made available to the computer system? How is the accuracy of the data verified? Will the system learn to apply sense rather than logic. For example if a small belligerently led nation, threatens a larger one with a nuclear missile attack, the logical computer will say fire our missiles first and destroy the threat and end the risk of attack. That is pure logic at work. Should we be governed by that?