Algorithmic aversion in artificial intelligence co-leadership and the impact of metaphors and comparisons

Abstract

The importance of Artificial Intelligence (AI) has been increasing at a fast pace. Besides transforming our lives, these technologies are crucial to improving business operations by conducting tasks faster, better and with lower costs, and by helping in the decision­making process that is an intellectually demanding task. However, despite the advantages AI can offer, people still disbelieve the ability of algorithms, often prefer to decide for themselves and refuse to rely on algorithms after seeing them err. This phenomenon is called algorithm aversion and goes against the best interest of companies that need to gain competitive advantage in a very competitive market. In this way, this dissertation intends to study the potential use of metaphors and comparisons as strategies to reduce algorithm aversion. For this to be done, an experimental study with three experimental groups was conducted. The effect of a language­based metaphor, a visual metaphor and an explicit comparison between human and artificial neurons was studied by relying on a specific type of AI called Artificial Neural Network (ANN) to see if people would prefer this technology that seems to function in a similar way to humans, over a general AI in a leadership position. The results of the study did not corroborate the hypothesis that the metaphors were going to reduce algorithm aversion, as the only difference found in the leadership acceptance was between the general AI group and the human one in which the new leader was a normal person.

Date of Award	27 Jan 2023
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Cristina Mendonça (Supervisor)