In the past few decades, artificial intelligence has gained an increasing amount of interest from the general public. Accompanying this interest, comes expectations of how sophisticated AI methods and their abilities are, often without a proper understanding of how they actually work. This demonstration is meant to give non-expert participants an idea of the view an RL agent has of its environment. We invite a volunteer to take the place of a standard RL agent and try learning the task solely based on information that would be available in a typical RL setting. The purpose of this demonstration is to illustrate how unintuitive an RL agent's perspective of its environments is from a human point of view, and hence how limited its understanding of the task it is learning is. By establishing this idea in non-experts minds, we hope to debunk certain inaccurate assumptions people may have about AI technologies, specifically RL in this case.
Plisnier, H, Fasano, AA & Nowe, A 2021, 'Play the Reinforcement Learning Agent', Paper presented at 33rd Benelux Conference on Artificial Intelligence and 30th Belgian-Dutch Conference on Machine Learning, Luxembourg, 10/11/21 - 12/11/21.
Plisnier, H., Fasano, A. A., & Nowe, A. (2021). Play the Reinforcement Learning Agent. Paper presented at 33rd Benelux Conference on Artificial Intelligence and 30th Belgian-Dutch Conference on Machine Learning, Luxembourg.
@conference{f2d62671a8b249e18320b0d986f037e6,
title = "Play the Reinforcement Learning Agent",
abstract = "In the past few decades, artificial intelligence has gained an increasing amount of interest from the general public. Accompanying this interest, comes expectations of how sophisticated AI methods and their abilities are, often without a proper understanding of how they actually work. This demonstration is meant to give non-expert participants an idea of the view an RL agent has of its environment. We invite a volunteer to take the place of a standard RL agent and try learning the task solely based on information that would be available in a typical RL setting. The purpose of this demonstration is to illustrate how unintuitive an RL agent's perspective of its environments is from a human point of view, and hence how limited its understanding of the task it is learning is. By establishing this idea in non-experts minds, we hope to debunk certain inaccurate assumptions people may have about AI technologies, specifically RL in this case.",
keywords = "Reinforcement Learning, Transparency, Volunteer-based Demonstration",
author = "Helene Plisnier and Fasano, {Alessandro Antonio} and Ann Nowe",
year = "2021",
month = nov,
day = "10",
language = "English",
note = "33rd Benelux Conference on Artificial Intelligence and 30th Belgian-Dutch Conference on Machine Learning : 33rd Benelux Conference on Artificial Intelligence and 30th Belgian-Dutch Conference on Machine Learning, BNAIC/BeneLearn 2021 ; Conference date: 10-11-2021 Through 12-11-2021",
url = "https://bnaic2021.uni.lu/",
}