To know what scares us we will first define AI. AI, short for artificial intelligence, is a very broad term which actually is not strictly defined. The easiest description is “intelligence demonstrated by machines” (Wikipedia, 2018). Within this broad definition there are three separations: narrow AI, general AI and super AI. The difference is that narrow AI is made to perform a single task, general AI understands and reasons with its environment as a human would, and super AI is smarter than humans in basically everything (Dickson, 2017). Currently the most development is in machine learning, which is a part of narrow AI. Machine learning refers to the process where a machine improves his own performance on a specific task, without humans explaining exactly how to accomplish that task (Brynjolfsson & McAfee, 2017). With machine learning we know the outcome, but we do not understand how the machine came to the answer. The learning process is mostly unknown to us.
We humans fear the unknown. Fear of the unknown can be described at its very basic “The perceived absence of information at any level of consciousness” (Carleton, 2016). Fear of the unknown is a fundamental fear, meaning that it is an emotion, continuously and normally distributed in the population, evolutionarily supported and irreducible (Carleton, Fear of the unknown: One fear to rule them all?, 2016).
And there we have our AI, solving tasks that we do not anymore understand how it has learned to solve them. Black boxes with data sets going that are too big for us to understand. We do not know anymore how that program has learned. As long as it works on task that we have given the AI we can handle it. But the moment that computers behave in a way that we did not predicted, that is the moment it gets scary for us. This is best illustrated by the two AI programs Facebook had, these programs were challenged to negotiate with each other over trade. Besides negotiating in the English language, the chatbots appeared to have developed an underlying language to communicate with each other (Griffin, 2017).
In the example of Facebook, the machines were only working on negotiating over trade. But this unknown behaviour already spikes our imagination. When we get to general AI the program will be able to apply its learning to other situations. Now think of the fact that we already have computers which, when working together, win war games from the best human players (Vincent, 2018). And what might happen if those computers start to have unknown behaviour.
Brynjolfsson, E., & McAfee, A. (2017). The Business of Artificial Intelligence. Boston: Harvard Business School Publishing Corporation.
Carleton, R. N. (2016). Fear of the unknown: One fear to rule them all? Regina: University of Regina.
Carleton, R. N. (2016). Into the Unknown: a review and synthesis of contemporary models involving uncertainty. Regina: University of Regina.
Dickson, B. (2017, 05 12). What is Narrow, General and Super Artificial Intelligence. Opgehaald van Techtalks: https://bdtechtalks.com/2017/05/12/what-is-narrow-general-and-super-artificial-intelligence/
Griffin, A. (2017, 07 31). Facebook’s artificial intelligence robots shut down after they start talking to each other in their own language. Opgehaald van Independent: https://www.independent.co.uk/life-style/gadgets-and-tech/news/facebook-artificial-intelligence-ai-chatbot-new-language-research-openai-google-a7869706.html
Vincent, J. (2018, 06 25). AI bots trained for 180 years a day to beat humans at Dota 2. Opgehaald van The Verge: https://www.theverge.com/2018/6/25/17492918/openai-dota-2-bot-ai-five-5v5-matches
Wikipedia. (2018, 10 12). Artificial intelligence. Opgehaald van Wikipedia: https://en.wikipedia.org/wiki/Artificial_intelligence
