Accountability and Reproducibility#

In brief#

Accountability and Reproducibility are two interrelated concepts, cornerstones of Trustworthy AI. Accoutable AI systems can contribute to reproducibility, and Reproducible AI systems can contribute to accountability.

More in detail#

Accountability and Reproducibility are two cornerstones of Trustworthy AI [1]. Accountability requires mechanisms be put in place to ensure that AI systems and their outcomes, both before and after their development, deployment and use, can be observed and analyzed. This ability to review AI systems involve technical and organisational logging processes [2] to enable investigators to draw the same conclusions from an experiment by following provided guidelines.

In this context, Accountability and Reproducibility are interrelated concepts. Developing reprodubicle AI systems can enable accountability over AI systems. On the other hand, the process of record-tracking and logging for accountability can support an increasing level of reproducibility.

A third dimension strictly correlated with Accountability and Reproducibility is Traceability. We suggest to navigate in the appropriate section of this book for more detailed information about these three dimensions.

Main Keywords#

  • Accountability: Accountability is an ethical aspect studied in the TAILOR project to ensure that a given actor or actors can render an account of the actions of an AI system. The accountability concept is strictly related to the concept of responsibility.

  • Wicked problems: A class of problems for which science provides insufficient or inappropriate resolution.

  • Meaningful human control: Meaningful human control is the notion that aims to generalize the traditional concept of operational control over technological artifacts to artificial intelligent systems. It implies that artificial systems should not make morally consequential decisions on their own, without appropriate control from responsible humans.

  • The Frame Problem: The frame problem is the challenge of knowing and modeling the relevant features and context of situations, and getting an agent to act on those without consideration all the irrelevant facts as well.

  • Reproducibility: Reproducibility is the ability of independent investigators to draw the same conclusions from an experiment by following the documentation shared by the original investigators.

  • Traceability: Traceability can be defined as the need to maintain a complete and clear documentation of the data, processes, artefacts and actors involved in the entire lifecycle of an AI model, starting from its design and ending with its production serving.

  • Provenance Tracking: Provenance tracking represents the tracking of “information that describes the production process of an end product, which can be anything from a piece of data to a physical object. […] Essentially, provenance can be seen as meta-data that, instead of describing data, describes a production process.”

  • Continuous Performance Monitoring: Continuous performance monitoring is the activity to track, log and monitor over time the behaviour and the performance of Artificial Intelligence and Machine Learning models. This activity is particularly relevant after in-production deployment in order to detect any performance drifts and outages of the model.

Bibliography#

1

European Commission, Content Directorate-General for Communications Networks, and Technology. Ethics guidelines for trustworthy AI. Publications Office, 2019. URL: https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai.

2

Jennifer Cobbe, Michelle Seng Ah Lee, and Jatinder Singh. Reviewable automated decision-making: a framework for accountable algorithmic systems. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, 598–609. 2021.

This entry was written by Luciano C Siebert.