Google’s AlphaGo AI has recently beaten the grandmaster of Go, Lee Se-Dol, 4-1. While this is a just a game, it’s clear that AI is starting to outsmart humanity. However, such defeat from an AI is not something new, with AI leaving chess players in the dust.
Additionally, AI found in self-driving cars are also already less accident prone than human drivers. So what makes the win by the AlphaGo AI so significant? Go has been regarded as a game requiring human intuition and pattern recognition. The fact that an AI could outsmart one of the world’s best players has ignited fears of artificial intelligence in the future.
Artificial intelligence should be considered a threat due to its ability to use ‘reinforcement learning’. This learning allows the computer to tweak its own strategy to learn the best moves from the outcome of its play. However, instead of playing against many humans, the computer plays against itself and hence it can play millions of games. In fact, AlphaGo played millions of games of Go before its match, clearly far more games than any human ever. This is a problem since the AI will explore many possible strategies that a human never would and hence every move that AlphaGo made was ‘not a human move’. The scope of AI is expanding to wider real-life applications and this is when AI could be problematic. For example, a stock-trading AI could (and most likely will) re-invent every method known to humanity on maximising return on investment and discover several that are unknown to us. These methods that the computer will find will be regarded as illegal and unethical price manipulation.
Ethics is the second part of the issue with artificial intelligence. Unfortunately, ethics and morality are not easily converted into rules and logic for a computer. While an AI may be able to imitate empathy and ethical behaviour, we cannot expect these forces to underpin its behaviour and decisions. This exploration of morality and ethics also shows that no matter how intellectually superior computers may become, they will never be able to replicate the human will to act in goodness and kindness no matter how logically unbeneficial it may be. In an age of what seems to be dehumanisation where technology is becoming more prevalent, the issues with AI may be a good reminder to us all on human traits such as generosity, humility, acceptance, respect and kindness. Perhaps then, humanity as a whole may be able to accept the differences between us and achieve peace to advance the progress of mankind.