Principles of Knowledge Construction in Artificial Intelligence Systems: Philosophical-Methodological Aspect

Introduction. The article addresses current philosophical problems of post-nonclassical science related to the analysis of machine learning-based intelligent systems. It contains a comparative analysis of constructivist and realistic approaches to identify the heuristic potential of radical constructivism in building a flexible functional architecture for AI.

Methodology and sources. The research is based on a comparative analysis of realistic and constructivist approaches. Its methodological apparatus includes the principles of operational closure and cognitive construction of reality, developed in the works of E. von Glasersfeld, J. Piaget, and their followers. The central thesis is the interpretation of knowledge not as a reflection of objective reality, but as a construction, the criterion of which is functional suitability for solving problems. This is reflected in modern machine learning concepts, such as reinforcement learning.

Results and discussion. It is demonstrated that the key principles of constructivism – the operational nature of knowledge, the iterative construction of cognitive structures, and the pragmatic criterion of viability – offer solutions to AI problems such as the “black box” problem, static nature of models, and contextual data dependency. This is achieved by rethinking machine learning as a process of active construction of functional representations and shifting the focus from accuracy to functional adequacy in specific applied contexts.

Conclusion. The developed tenets of radical constructivism help to rethink the nature of data and models in machine learning and open up prospects for a deeper analysis of the requirements for iterative and adaptive learning architectures and the development of artificial intelligence systems.

1. Sergeev, S.F. (2023), “Postclassical Science: Recursion of Meaning in Network Language Models”, Tret'i Stepinskie chteniya. Perspektivy filosofii nauki v sovremennuyu epokhu [The Third Stepinsky Readings. Prospects for the Philosophy of Science in the Modern Era], Int. Conf., Moscow, RUS, 20–21 June 2023, pp. 390–396.

2. Semenov, A. (2017), “Seymour Papert and Us. Constructionism as the Educational Philosophy of the 21st Century”, Educational Studies. Moscow, no. 1, рp. 269–294. DOI: 10.17323/1814-9545-2017-1-269-294

3. Weintrop, D. and Wilensky, U. (2017), “Between a block and a typeface: Designing and evaluating hybrid programming environments”, IDC '17: Proc. of the 2017 Conf. on Interaction Design and Children, pp. 183–192. DOI: 10.1145/3078072.3079715

4. Von Glasersfeld, E. (1984), “An introduction to radical constructivism”, The invented reality: how do we know what we believe we know? Contributions to constructivism, in Watzlawick, P. (ed.), W. W. Norton & Co., NY, USA, pp. 17–40.

5. Baksanskii, O.E. (2022), “Radical constructivism and knowledge”, Bol'shaya Evraziya: razvitie, bezopasnost', sotrudnichestvo [Greater Eurasia: Development, Security, Cooperation], no. 5-1, pp. 1000–1002.

6. Kasavin, I.T. (2008), “Constructivism: declared programs and unsolved problems”, Epistemology & Philosophy of Science, vol. XV, pp. 5–14.

7. Von Glasersfeld, E. (1989), “Cognition, construction of knowledge, and teaching”, Synthese, vol. 80, no. 1, pp. 121–140.

8. Yanchenko, A.A. (2014), “Methodological potential of epistemological constructivism for the development of artificial intelligence theory”, Istoricheskie, filosofskie, politicheskie i yuridicheskie nauki, kul'turologiya i iskusstvovedenie. Voprosy teorii i praktiki [Historical, Philosophical, Political and Law Sciences, Culturology and Study of Art. Issues of Theory and Practice], no. 1 (39), part 2, pp. 210–215.

9. Johnston, J.W. (2023), “The Construction of Reality in an AI: A Review”, Arxiv.org, available at: https://arxiv.org/pdf/2302.05448 (accessed 17.08.2025). DOI: https://doi.org/10.48550/arXiv.2302.05448

10. Bishop, C.M. (2006), Pattern recognition and machine learning, Springer, NY, USA.

11. Chernikova, I.V. and Chernikova, D.V. (2019), “Methodological and structural transformations of the contemporary science“, Tomsk State Univ. J. of Philosophy, Sociology and Political Science, no. 49, pp. 60–68. DOI: 10.17223/1998863X/49/7

12. Nowak, M., Castellini, C. and Massironi, C. (2018), “Applying radical constructivism to machine learning: a pilot study in assistive robotics”, Constructivist Foundations, vol. 13, no. 2, pp. 250–262.

13. Li, X. and Huan, J. (2017), “Constructivism learning: A learning paradigm for transparent predictive analytics”, Proc. of the 23rd ACM SIGKDD Int. Conf. on Knowledge Discovery and Data Mining, Halifax, Canada, 13–17 Aug. 2017, pp. 285–294. DOI: http://dx.doi.org/10.1145/3097983.3097994

14. Mathewson, K.W. and Pilarski, P.M. (2017), “Reinforcement learning based embodied agents modelling human users through interaction and multi-sensory perception”, AAAI spring symposium on interactive multi-sensory object perception for embodied agents, Stanford, USA, 27–29 March 2017, pp. 477–481. DOI: 10.48550/arXiv.1701.02369

15. Sarkar, A. (2016), “Constructivist design for interactive machine learning”, CHI EA '16: Proc. of the 2016 CHI Conf. Extended Abstracts on Human Factors in Computing Systems, San Jose, USA, 07–12 May 2016, pp. 1467–1475. DOI: https://doi.org/10.1145/2851581.2892547

16. Fails, J.A. and Olsen, D.R. (2003), “Interactive machine learning”, IUI ’03: Proc. of the 8th Int. Conf. on Intelligent user Interfaces, Miami, USA, 12–15 Jan. 2003, pp. 39–45. DOI: 10.1145/604050.604056

17. Tokuhama-Espinosa, T. and Borja, C. (2023), “Radical neuroconstructivism: a framework to combine the how and what of teaching and learning?”, Frontiers in Education, vol. 8: 1215510. DOI: 10.3389/feduc.2023.1215510

18. Abrahamson, D., Shayan, A., Bakker, S. et al. (2016), “Exposing Piaget's scheme: Empirical evidence for the ontogenesis of coordination in learning a mathematical concept”, ICLS 2016: Proc. of the 12th Int. Conf. of the Learning Sciences, Singapore, SGP, 20–24 June 2016, pp. 466–473.