Aim of AI - Artificial Intelligence

A renown example of the power of artificial intelligence was when the machine beat the breakout game in a rather unexpected way. The game is simple. Akin to tennis you have a pad to hit a ball up the field and try and take out the blocks at the top, one at a time. Bit by bit the wall is eroded as you keep returning the ball. It gets harder as the angles change and the speed increases. The AI discovered a flaw in the game.

If it gets the ball behind the wall, the ball will bounce fast and furiously removing the blocks without any need to keep returning the ball. The AI found a way to get the highest score in the quickest most efficient way. This brings me to argue that AI has a fundamental problem. What is the aim? Do we want machines to work in the most productive, most effective way possible to bring about a certain result. In this case it is the highest score. In other areas it might be to maximise profits or cause the most amount of damage to an enemy. We as humans set the bar, set the goal and define what we want the machine to do. Maybe that will change over time and the machine will create a new future that humans never considered. But will it be a world that suits us? Will it be a world that excludes the minority and focuses on the will of the majority.

Radio stations that play tracks that are the most popular are set to keep the maximum number of listeners tuned in. It works to some degree, but how dull and dreary, how boring it soon becomes. We assume that if the algorithm is producing the most it can that it is a good thing. What though do we want to produce. The greatest output for the least input. No room for pleasure and real human contentment. Will AI lead to progress of sorts that is devoid of that thing we are not supposed to have, namely, fun?

More thought on this is required. Am I being too romantic and yearning for the nostalgia of the past when things were slower and more haphazard and fail to see the merits of this goal to go faster, cheaper, and more accurately.

Can the aim of AI be flawed if we don't recognise what philosophical end goals we want to achieve?

Aim