It is said that AI will revolutionize the world around us and the way we live. Yes, this is probably very true. From how we do groceries, to our job, ending with to who we marry. After the polemics on the elections of Donald Trump as USA president, it seems that also the politic systems are not immune to the super power of artificial intelligence. It is indeed easy to imagine the repercussions of the use of AI and machine learning in the political scene. AI algorithms can influence vote fraud, disrupt the way propaganda is currently made by politicians, and hugely affect decision making processes in the context of fiscal and monetary policy.
Yet, everybody is so focused on the impact that AI can have on people that we tend to forget the impact that some people can have on AI. And it is on this ‘some’ that I would like the attention to be focused, as most likely I will not have any influence on super intelligent machines, but governments surely will. In a holistic consideration of how AI is created and programmed, governments will have a say. A big one. Both for power interest and security reasons, States cannot leave super advanced technologies take over without their supervision and, more precisely, their almost total control. As a proof, government and intelligence agencies are already closely monitoring AI developments. Indeed, if AI is powerful, Governments are too, and stakes are too high for them to stay out of it. For that reason, chances are that any important AI project will either be directly funded by the State or belong to an institution closely linked to it. And, even if it wouldn’t be the case, Government entities can have influence on basically anything one can think about. Now, this wouldn’t be a problem if politicians were all honest and caring about citizens’ good. But, looking back at both old and modern history, we all know that this is true in theory but not in practice.
Let’s switch the direction of our logic and try to think about the influence that politics may have on AI, instead of the way around. We are so busy wondering whether AI will be a friendly or unfriendly entity that we forget that AI will not program itself. And as much as programmers can identify themselves as independent thinkers, there’s no way they can avoid being influenced, consciously or subconsciously, by political climate and ideologies. Thus, friendliness of AI towards the human kind is a political and institutional problem as much as an algorithmic one. How would you secure AI fairness when its roots are not?
References
Brannen, R. (2016). Politics Is Upstream of AI. [online] The Future Primaeval. Available at: https://thefutureprimaeval.net/politics-is-upstream-of-ai/ [Accessed 14 Oct. 2018].
