Feier et al. (2022) – Hiding behind machines: Artificial agents may help to evade punishmen

Question 1

Abstract

Answer 1

The transfer of tasks with sometimes far-reaching implications to autonomous sys- tems raises a number of ethical questions. In addition to fundamental questions about the moral agency of these systems, behavioral issues arise. We investigate the empirically accessible question of whether the imposition of harm by an agent is systematically judged differently when the agent is artificial and not human. The results of a laboratory experiment suggest that decision-makers can actually avoid punishment more easily by delegating to machines than by delegating to other people. Our results imply that the availability of artificial agents could provide stronger incentives for decision-makers to delegate sensitive decisions.

Question 2

Blame avoidance

Answer 2

The concept of blame and the role of empirical research
* In which sense can artificial agents be held responsible for their actions?
* Blame: reaction to something of negative normative significance about someone or
their behavior
* Retribution gap: humans are innate retributivists who seek to identify and punish
culpable wrongdoers whenever they perceive harm (it exists and affects behavior)
- This becomes problematic when harm is brought about by robots or other non-human
agents
- Example: self-driving cars

Question 3

Results

Answer 3

Results
* Delegators were rewarded equally for good outcomes that were caused by either human
or artificial agents
* Delegators did not effectively avoid punishment for a bad outcome if their human agent
caused it instead of themselves. However, punishment was significantly lower if their
artificial agent caused it instead of themselves
* Participants who were more at risk were also more likely to delegate
* It appears that the expected utility of delegating was the driving factor behind participants’
decisions to do so or not, regardless of the nature of their agent
* Those who were less confident regarding their own performance were more likely to
delegate the task
* Subjects who performed better in the logic task were accordingly less likely to delegate
* Machine agents can be successfully used to avoid punishmen