What does accountability actually mean, and how does it apply to AI ethics? We’ll also discuss what moral agency and responsibility mean and the difficulty of assigning blame.

I. Algorithms and accountability

Parking control services provide an example of how algorithms are increasingly used for automating public services. As algorithms are exact, fast and precise, they often promote better service efficiency, reliability and consistency. Paradoxically, algorithms can also make systematic errors, be biased and cause serious harms. For example, scanning systems may malfunction, or suffer from bugs. They may make mistakes and suggest the tickets be issued on invalid grounds. In these cases, who should take the responsibility – and on what grounds?

Although we say things like "yes, it was the algorithm’s fault and it is responsible for the wrong decision", we do not literally mean that contemporary algorithms would be morally guilty. Instead, the algorithms are causal factors that underlie the decisions. Mere causes, however, differ from morally responsible actions.

Even though algorithms themselves cannot be held accountable as they are not moral or legal agents, the organizations designing and deploying algorithms can be taken to be morally responsible through governance structures. Thus, in the case of the city of Amsterdam, it is the human inspector that makes the final decision – and also takes responsibility. However, one day the human inspector may be replaced by algorithms, too. Who, then, will take responsibility?

You have reached the end of this section!

Continue to the next section: