The application of artificial intelligence in business raises the question of managerial responsibility: to what extent are managers held responsible for the ‘automated’ decisions taken by the tools entrusted to them?
Automated decision-making leverages algorithms based on derived profiles, as well as assumptions about how we think.
The interpretability of these ‘black boxes’ is a real challenge – because humans, even those who design them, cannot understand how the variables are combined to make predictions.
The value of AI in decision-making doesn’t come from the technology itself, but from how organizations leverage AI to enhance the processes of managerial decision-making.
In the articles reviewed below, the authors discuss the mechanics of how AI can support managerial decision-making, the challenges of capturing the context of the decision environment, and the ethical questions that automated decision-making brings to light.
The authors point out the various potential use cases of AI assisted decision-making in business and society, the potential value that AI produces, as well as the implicit ethical and legal questions that have been raised.
They argue, from a variety of angles, that AI’s benefits will be conditioned by how organizations manage the practical issues of the implementation and supervision of such systems.
Enjoy reading!
Article 1:
AI Ethical Decision Making:
Is Society Ready?
This summary reviews research at Hiroshima University addressing the question of whether society is willing to accept AI decision-making. In their multi-country study of human interaction with autonomous cars.
The researchers ran two experiments, the first designed to measure the bias people might have against AI ethical decision-making, and the second to test the effect of two alternative ways of introducing AI into society.
They find that people do not seem to show bias against AI ethical decision-making but rather an aversion to AI. The extent of this bias varies both with national culture and the degree of economic development.
Source:
ScienceDaily & Hiroshima University
Johann Caro-Burnett & Shinji Kaneko
Article 2:
Algorithmic Decision Making:
To What Extent Can Due Process Principles Assist in Protecting Individuals from the Threat of Algocracy
Michael Camilleri’s dissertation at the University of Edinburgh explores the degree to which traditional mechanisms of government can integrate algorithmic decision-making without violating due process principles.
The author argues that the crucial question here is whether the legislature should allow the courts to tackle the ‘unique’ aspects of these threats or whether it is necessary to legislate.
He concludes that this dissertation highlights the threats of algocracy not to argue that the use of algorithmic decision-making systems by public bodies should be barred but acknowledges that there are ethical and legal problems that need addressing.
Source:
Michael Camilleri
Article 3:
From Rule of Law to Statute Drafting:
Legal Issues for Algorithms in Government Decision-Making
In this contribution to the Cambridge Handbook on the Law of Algorithms, Monika Zalnieriute and Lyria Bennett Moses examine how the values of the rule of law are affected by automated government decision-making systems and the legal and practical issues that the implementation and supervision of such systems may pose in practice.
The chapter sketches the means with which algorithms are used across the spectrum of government decision-making — from the drafting of legislation, to judicial decision-making.
Source:
Monika Zalnieriute and Lyria Bennett Moses
Article 4:
AI Based Decision Making:
Combining Strategies to Improve Operational Performance
A. Al-Surmi, M. Bashiri, and I. Koliousis provide decision-makers with a deep learning-based model that leverages artificial neural networks (ANN), to improve decision performance using marketing and IT strategy.
Towards this end, the researchers used strategic decision-making theory, which involves structural equation modeling (SEM) and the use of an artificial neural network (ANN) algorithm.
The paper details the researchers’ hypotheses and conclusions. The article begins with the details of the theoretical research and then their research into existing algorithms and those they have implemented.
Much more than the provision of a new algorithm, the researchers provide a theoretical, methodological and empirical contribution to understanding AI’s potential role in organizational decision-making.
Source:
A. Al-Surmi, M. Bashiri, and I. Koliousis
Article 5:
Artificial Intelligence isn’t Ready to Make Unsupervised Decisions
In this contribution to HBR; Joe McKendrick and Andy Thurai use a myriad of examples to argue that AI in its current form misses the larger context which shapes the decision-making process and thus can’t analyze its own decisions with reasoning behind it.
Although research in AI has evolved to compete with human intelligence in many areas with unmatched accuracy, quality, and speed, there is little evidence that AI will be able to capture and process the subjective experiences, feelings, and empathy that makes the world a better place to live and work.
Source:
Joe McKendrick and Andy Thurai
Article 6:
What AI-Driven Decision Making Looks Like?
In this HBR contribution, Eric Colson highlights the distinction between “data-driven” and “AI-driven“.
The author points out that despite the information gained through AI (with varying degrees of relevance in the results), humans have difficulties (cognitive biases) in processing the results.
Although he advocates the increasing use of AI in the decision-making process, he warns against the risk of misuse of biased data and the absence of humans in the loop to ensure that company values and market dynamics are taken into account.
Source:
Eric Colson
Article 7:
Will Algorithms Blind People?
The Effect of Explainable AI and Decision-makers’ Experience on AI-supported Decision-making in Government
In this HBR contribution, Eric Colson highlights the distinction between “data-driven” and “AI-driven“.
The author points out that despite the information gained through AI (with varying degrees of relevance in the results), humans have difficulties (cognitive biases) in processing the results.
Although he advocates the increasing use of AI in the decision-making process, he warns against the risk of misuse of biased data and the absence of humans in the loop to ensure that company values and market dynamics are taken into account.
On the other, organizations need to actively work to ensure the transparency and accountability of decision-making.
Source:
SAGE Journals
Janssen, M., Hartog, M., Matheus, R., Yi Ding, A., & Kuk, G.

