Ethical AI Starts with Data Provenance: A FATE Perspective
DOI:
https://doi.org/10.15680/553fyn65Keywords:
Ethical AI, Data Provenance, Fairness, Accountability, Transparency, Ethics, FATE, AI Governance, Bias Mitigation, Explainable AI, Data Lineage, Responsible AI, AI TransparencyAbstract
With artificial intelligence (AI) becoming increasingly integral to decision-making in various sectors, ensuring its ethical deployment has never been more critical. Central to fostering ethical AI is the concept of data provenance—the ability to trace and understand the origins, transformations, and final usage of data. Data provenance plays a pivotal role in promoting Fairness, Accountability, Transparency, and Ethics (FATE) in AI systems. This paper explores the critical role of data provenance in AI, discusses how it can be leveraged to uphold FATE principles, and examines methodologies for implementing effective data lineage systems. By ensuring that data is traceable, auditable, and ethically sourced, data provenance forms the foundation of responsible AI deployment, helping mitigate risks such as bias, discrimination, and opacity in AI decision-making processes.
References
1. Binns, R. The ethics of explainable AI. Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems.
2. Dastin, J. ). Amazon scraps secret AI recruiting tool that showed bias against women. Reuters.
3. Mehrabi, N., et alA survey on bias and fairness in machine learning. ACM Computing Surveys, 52(4), 1–35.
4. Narayanan, A., et al. ). 18.3. Fairness in machine learning: A survey. Proceedings of the 2018 ACM/IEEE Symposium on Edge Computing.
5. Sculley, D., et al. Hidden Technical Debt in Machine Learning Systems. Neural Information Processing Systems (NeurIPS).
6. Voigt, P., & Von dem Bussche, A.The EU General Data Protection Regulation (GDPR): A Practical Guide. Springer.
7. European Commission. Artificial Intelligence Act – Proposal for Regulation of the European Parliament.
8. Moreau, L., et al. The Open Provenance Model Core Specification. Future Generation Computer Systems, 27(6), 743–756.
9. Apache Atlas. https://atlas.apache.org.
10. OpenLineage. https://openlineage.io.
11. MLflow. (2023). https://mlflow.org.
12. Pachyderm. (2024). https://www.pachyderm.io.
13. Comet ML. (2023). https://www.comet.com.
