Legal Means of Providing the Principle of Transparency of the Artificial Intelligence

Objective: to analyze the current technological and legal theories in order to define the content of the transparency principle of the artificial intelligence functioning from the viewpoint of legal regulation, choice of applicable means of legal regulation, and establishing objective limits to legal intervention into the technological sphere through regulatory impact.

Methods: the methodological basis of the research is the set of general scientific (analysis, synthesis, induction, deduction) and specific legal (historical-legal, formal-legal, comparative-legal) methods of scientific cognition.

Results: the author critically analyzed the norms and proposals for normative formalization of the artificial intelligence transparency principle from the viewpoint of impossibility to obtain the full technological transparency of artificial intelligence. It is proposed to discuss the variants of managing algorithmic transparency and accountability based on the analysis of social, technical and regulatory problems created by algorithmic systems of artificial intelligence. It is proved that transparency is an indispensible condition to recognize artificial intelligence as trustworthy. It is proved that transparency and explainability of the artificial intelligence technology is essential not only for personal data protection, but also in other situations of automated data processing, when, in order to make a decision, the technological data lacking in the input information are taken from open sources, including those not having the status of a personal data storage. It is proposed to legislatively stipulate the obligatory audit and to introduce a standard, stipulating a compromise between the technology abilities and advantages, accuracy and explainability of its result, and the rights of the participants of civil relations. Introduction of certification of the artificial intelligence models, obligatory for application, will solve the issues of liability of the subjects obliged to apply such systems. In the context of professional liability of professional subjects, such as doctors, militants, or corporate executives of a juridical person, it is necessary to restrict the obligatory application of artificial intelligence if sufficient transparency is not provided.

Scientific novelty: the interdisciplinary character of the research allowed revealing the impossibility and groundlessness of the requirements to completely disclose the source code or architecture of the artificial intelligence models. The principle of artificial intelligence transparency may be satisfied through elaboration and provision of the right of the data subject and the subject, to whom the decision made as a result of automated data processing is addressed, to reject using automated data processing in decision-making, and the right to object to the decisions made in such a way.

Practical significance: is due to the actual absence of sufficient regulation of the principle of transparency of artificial intelligence and results of its functioning, as well as the content and features of the implementation of the right to explanation the right to objection of the decision subject. The most fruitful way to establish trust towards artificial intelligence is to recognize this technology as a part of a complex sociotechnical system, which mediates trust, and to improve the reliability of these systems. The main provisions and conclusions of the research can be used to improve the legal mechanism of providing transparency of the artificial intelligence models applied in state governance and business.

1. Balduzzi, D., Frean, M., Leary, L., Lewis, J. P., Ma, K. W. D., & McWilliams, B. (2017). The shattered gradients problem: If resnets are the answer, then what is the question? In International Conference on Machine Learning (pp. 342–350). PMLR.

2. Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? In Proceedings of the 2021 ACM conference on fairness, accountability, and transparency (pp. 610–623).

3. Buolamwini, J., & Gebru, T. (2018). Gender shades: Intersectional accuracy disparities in commercial gender classification. In Conference on fairness, accountability and transparency (pp. 77–91). PMLR.

4. Camilleri, M. A. (2018). Market segmentation, targeting and positioning. In Travel marketing, tourism economics and the airline product (pp. 69–83). New York: Springer.

5. Cho, Y. H., Kim, J. K., & Kim, S. H. (2002). A personalized recommender system based on web usage mining and decision tree induction. Expert systems with Applications, 23(3), 329–342. https://doi.org/10.1016/s0957-4174(02)00052-0

6. Cragun, B. J., & Steudel, H. J. (1987). A decision-table-based processor for checking completeness and consistency in rule-based expert systems. International Journal of Man-Machine studies, 26(5), 633–648. https://doi.org/10.1016/s0020-7373(87)80076-7

7. Dyakonova, M. O., Efremov, A. A., Zaitsev, O. A., et al.; I. I. Kucherova, S. A. Sinitsyna (Eds.). (2022). Digital economy: topical areas of legal regulation: scientific-practical tutorial. Moscow: IZISP, NORMA. (In Russ.).

8. Foss-Solbrekk, K. (2021). Three routes to protecting AI systems and their algorithms under IP law: The good, the bad and the ugly. Journal of Intellectual Property Law & Practice, 16(3), 247–258. https://doi.org/10.1093/jiplp/jpab033

9. Goodfellow, I. J., Shlens, J., & Szegedy, C. (2014). Explaining and harnessing adversarial examples. arXiv preprint arXiv:1412.6572.

10. Goodman, B., & Flaxman, S. (2017). European Union regulations on algorithmic decision-making and a “right to explanation”. AI magazine, 38(3), 50–57. https://doi.org/10.1609/aimag.v38i3.2741

11. Gunning, D. (2017). Explainable artificial intelligence (XAI). Defense advanced research projects agency (DARPA). nd Web. 2(2), 1.

12. Hacker, P., Krestel, R., Grundmann, S., & Naumann, F. (2020). Explainable AI under contract and tort law: legal incentives and technical challenges. Artificial Intelligence and Law, 28(4), 415–439. https://doi.org/10.1007/s10506-020-09260-6

13. Hastie, T., Tibshirani, R., Friedman, J. H., & Friedman, J. H. (2009). The elements of statistical learning: data mining, inference, and prediction (Vol. 2). New York: Springer.

14. Hofmann, T., Schölkopf, B., & Smola, A. J. (2008). Kernel methods in machine learning.

15. Kalis, B., Collier, M., & Fu, R. (2014). 10 promising AI applications in health care. Harvard Business Review.

16. Kalpokas, I. (2019). Algorithmic Governance. Politics and Law in the Post-Human Era. Cham: Palgrave Pivot.

17. Kalyatin, V. O. (2022). Deepfake as a legal problem: new threats or new opportunities? Zakon, 7, 87–103. (In Russ.). https://doi.org/10.37239/0869-4400-2022-19-7-87-103

18. Kharitonova, Y. S., Savina, V. S., & Pagnini, F. (2021). Artificial Intelligence’s Algorithmic Bias: Ethical and Legal Issues. Perm U. Herald Jurid. Sci, 3(53), 488–515. https://doi.org/10.17072/1995-4190-2021-53-488-515

19. Koene, A., Clifton, C., Hatada, Y., Webb, H., & Richardson, R. (2019). A governance framework for algorithmic accountability and transparency. Panel for the Future of Science and Technology.

20. Kuteynikov, D. L., Izhaev, O. A., Zenin, S. S., & Lebedev, V. A. (2020). Algorithmic Transparency and Accountability: Legal Approaches to Solving the “Black Box” Problem. Lex Russica, 73(6), 139–148. (In Russ.). https://doi.org/10.17803/1729-5920.2020.163.6.139-148

21. Lawrence, R. L., & Wright, A. (2001). Rule-based classification systems using classification and regression tree (CART) analysis. Photogrammetric engineering and remote sensing, 67(10), 1137–1142.

22. LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436–444.

23. Maggiolino, M. (2018). EU trade secrets law and algorithmic transparency. Available at SSRN 3363178.

24. Malgieri, G., & Comandé, G. (2017). Why a right to legibility of automated decision-making exists in the general data protection regulation. International Data Privacy Law, 7(4), 243–265. https://doi.org/10.1093/idpl/ipx019

25. McEwen, R., Eldridge, J., & Caruso, D. (2018). Differential or deferential to media? The effect of prejudicial publicity on judge or jury. The International Journal of Evidence & Proof, 22(2), 124–143. https://doi.org/10.1177/1365712718765548

26. Najafabadi, M. M., Villanustre, F., Khoshgoftaar, T. M., Seliya, N., Wald, R., & Muharemagic, E. (2015). Deep learning applications and challenges in big data analytics. Journal of Big Data, 2(1), 1–21. https://doi.org/10.1186/s40537-014-0007-7

27. Pak, M., & Kim, S. (2017). A review of deep learning in image recognition. In 2017 4th international conference on computer applications and information processing technology (CAIPT). https://doi.org/10.1109/caipt.2017.8320684

28. Payre, W., Cestac, J., & Delhomme, P. (2014). Intention to use a fully automated car: Attitudes and a priori acceptability. Transportation research part F: traffic psychology and behavior, 27, 252–263.

29. https://doi.org/10.1016/j.trf.2014.04.009

30. Plous, S. E. (2003). Understanding prejudice and discrimination. McGraw-Hill.

31. Quillian, L. (2006). New approaches to understanding racial prejudice and discrimination. Annu. Rev. Sociol., 32, 299–328. https://doi.org/10.1146/annurev.soc.32.061604.123132

32. Samek, W., Montavon, G., Lapuschkin, S., Anders, C. J., & Müller, K. R. (2021). Explaining deep neural networks and beyond: A review of methods and applications. Proceedings of the IEEE, 109(3), 247–278. https://doi.org/10.1109/jproc.2021.3060483

33. Savelyev, A. I. (2021). Scientific-practical article-by-article commentary to Federal Law “On personal data” (2nd ed., amended and abridged). Moscow: Statut. (In Russ.).

34. Selbst, A., & Powles, J. (2017). “Meaningful information” and the right to explanation. International Data Privacy Law, 7(4), 233–242. https://doi.org/10.1093/idpl/ipx022

35. Silkin, V. V. (2021). Transparency of executive power in digital epoch. Russian Juridical Journal, 4, 20–31. (In Russ.). https://doi.org/10.34076/20713797_2021_4_20

36. Silver, D. et al. (2017). Mastering the game of go without human knowledge. Nature, 550(7676), 354.

37. Vargas, R., Mosavi, A., & Ruiz, R. (2018). Deep learning: a review. Preprints.org. https://doi.org/10.20944/preprints201810.0218.v1

38. von Eschenbach, W. J. (2021). Transparency and the black box problem: Why we do not trust AI. Philosophy & Technology, 34(4), 1607–1622.

39. Wachter, S., Mittelstadt, B., & Floridi, L. (2017). Why a right to explanation of automated decision-making does not exist in the general data protection regulation. International Data Privacy Law, 7(2), 76–99. https://doi.org/10.1093/idpl/ipx005

40. Wulf, A. J. & Seizov, O. (2022). “Please understand we cannot provide further information”: evaluating content and transparency of GDPR-mandated AI disclosures. AI & SOCIETY, 1–22. https://doi.org/10.1007/s00146022-01424-z

41. Yampolskiy, R. V. (2019). Unexplainability and incomprehensibility of artificial intelligence. arXiv preprint arXiv:1907.03869

42. 许可、刘畅. 论算法备案制度 // 人工智能. 2022. № 1. P. 66. [Xu Ke, Liu Chang. (2022). On the Algorithm Filing System. Artificial Intelligence, 1, 66.]