Interpretazione e comprensione: è possibile una lettura ermeneutica delle interazioni umano-LLM?
DOI:
https://doi.org/10.60923/issn.2532-8816/25598Parole chiave:
Ermeneutica, grandi modelli linguistici, Prompting, Responsabilità, AIUCD2025Abstract
Il contributo propone una lettura ermeneutica delle interazioni tra esseri umani e modelli linguistici. La tesi di fondo è che un modello linguistico non è un soggetto che comprende nel senso forte dell’ermeneutica classica, ma un sistema statistico che traduce la precomprensione (Vorverständnis) degli utenti in vincoli di pertinenza, senza produrre un’autonomia ermeneutica anche laddove eccedano l’intenzione iniziale. Richiamando pensatori della tradizione ermeneutica quali Droysen, Dilthey e Gadamer, in particolare la concezione di “costituzione della domanda” e il concetto gadameriano di “gioco”, il paper analizza il prompt come atto performativo, distinguendo diverse forme di opacità (computazionale, epistemica, ermeneutica) e diversi regimi della relazione umano-modello. Questo schema di un’ermeneutica filosofica applicata all’AI intende integrarsi con le analisi critiche dell’infrastruttura al fine di sviluppare pratiche di responsabilità interpretativa.
Riferimenti bibliografici
Analecta Hermeneutica. 2023. The Hermeneutics of Artificial Intelligence. Analecta Hermeneutica, 15(1).
Arora, A., Jurafsky, D., Moore, J., Ong, D., et. al. 2026. Verbalizing LLMs’ assumptions to explain and control sycophancy. ArXiv:2604.03058v3.
Belcak, P., et al. 2025. Small Language Models are the Future of Agentic AI. ArXiv:2506.02153
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., et al. 2020.
Language models are few-shot learners. Advances in Neural Information Processing Systems, 33, 1877–1901. Preprint version: arXiv:2005.14165.
Bentley, M. 1999. Modern Historiography. An Introduction. London: Routledge.
Besta, M., et al. 2024. Graph of Thoughts: Improving Large Language Model Reasoning with Graph-Based Prompting. Proceedings of the AAAI Conference on Artificial Intelligence. ArXiv:2308.09687
Castelvecchi, D. 2016. Can we open the black box of AI? In Nature 538, n. 7623: 2, 0–23. doi:10.1038/538020a.
Churchland, P. M. 1989. A Neurocomputational Perspective: The Nature of Mind and the Structure of Science. Cambridge, MA: MIT Press.
Coeckelbergh, M. 2023. Narrative responsibility and artificial intelligence. In AI & Society, 38, 2437–2450. https://doi.org/10.1007/s00146-021-01375-x
Collingwood, R. G. 1946. The Idea of History. Oxford: Clarendon Press.
Dennett, D. C. 1987. The Intentional Stance. Cambridge, MA: MIT Press.
Dilthey, W. 1883. Einleitung in die Geisteswissenschaften. Leipzig: B. G. Teubner.
Dikici, S., Bilgin, T. T.. 2026. Small Language Models: A Systematic Review of Computational Trade-Offs, Privacy Advantages and Deployment in Intelligent Systems. Expert Systems 43.8: e70331. https://doi.org/10.1111/exsy.70331.
Doshi Velez, F., and Been, K. 2017. “Towards a Rigorous Science of Interpretable Machine Learning.” arXiv:1702.08608.
Droysen, J. G. 1882. Grundriss der Historik (3rd ed.). Leipzig: Veit & Comp.
Fan, L., et al. 2022. MineDojo: Building Open-Ended Embodied Agents with Internet-Scale Knowledge. NeurIPS 2022 Datasets and Benchmarks ArXiv:2206.08853.
Feyerabend, P. 1993. Against Method. 3rd ed. London–New York: Verso.
Floridi, L. 2025. AI as Agency without Intelligence: On Artificial Intelligenceas a New Form of Artificial Agency and the MultipleRealisability of Agency Thesis. Philosophy & Technology (2025) 38:30 https://doi.org/10.1007/s13347-025-00858-9
Foucault, M. 1971. L’ordre du discours. Leçon inaugurale au Collège de France prononcée le 2 décembre 1970. Paris: Gallimard.
Frankish, K. 2024. What are large language models doing? In A. Strasser (Ed.), Anna’s AI Anthology: How to Live with Smart Machines?, 55–78. Xenomoi.
Gadamer, H. G. 1990. Hermeneutik I. Wahrheit und Methode: Grundzüge einer philosophischen Hermeneutik. Gesammelte Werke, Bd. 1. Tübingen: J. C. B. Mohr (Paul Siebeck).
Gao, C., et al. 2024. EmbodiedCity: A Benchmark Platform for Embodied Agent in Real-world City Environment. ArXiv:2410.09604.
Gudibande, A., et al. 2024. The False Promise of Imitating Proprietary Language Models. ICLR 2024 Conference. Versione OpenReview 2024: arXiv:2305.15717.
Habermas, J. 1989. Etica del discorso. A cura di E. Agazzi. Roma-Bari, Laterza.
Hegel, G. W. F. 1986. Grundlinien der Philosophie des Rechts. Gesammelte Werke, Bd. 14. Hamburg: Meiner.
Heidegger, M. 2005 [1927]. Essere e tempo. A cura di F. Volpi su tr. it. P. Chiodi. Milano, Longanesi.
Hempel, C. G. 1965. Aspects of scientific explanation. New York, Free Press.
Hooke, J. D. F. 2023. Martin Heidegger’s concept of understanding (Verstehen): An inquiry into artificial intelligence. In Analecta Hermeneutica, 15(1), 138–159. Special Issue: The Hermeneutics of Artificial Intelligence.
Humphreys, P. 2004. Extending Ourselves: Computational Science, Empiricism, and Scientific Method. Oxford: Oxford University Press.
Husserl, E. 2025. Ideen zu einer reinen Phänomenologie und phänomenologischen Philosophie. Zweites Buch: Phänomenologische Untersuchungen zur Konstitution und Wissenschaftstheorie sowie das Nachwort zu meinen Ideen. Hrsg. von Dirk Fonfara. Cham: Springer (Husserliana, neue Edition).
Li Z. et al.. 2025, AgentFlow: Modulating LLM Agents via Workflow Generation, https://zhuofeng-li.github.io/
Lipton, Z. C. 2018. The Mythos of Model Interpretability. In Queue 16 (3): 31–57.
Maohao, R., et al. 2026. CaveAgent: Transforming LLMs into Stateful Runtime Operators. ArXiv:2601.01569
Marx, K. 1974. Tesi su Feuerbach. In L’ideologia tedesca, 2ª ed., 15–18. Roma: Editori Riuniti.
O’Hara, K. 2020. Explainable AI and the philosophy and practice of explanation. In Computer Law & Security Review, 39, 1-7.
Pasquinelli, M. 2023. The Eye of the Master: A Social History of Artificial Intelligence. London: Verso Books.
Qi X., et al. 2024. Fine Tuning Aligned Language Models Compromises Safety. ICLR 2024 Conference. ArXiv:2310.03693
Rawat A., et al. 2025. Pre Act: Multi step Planning and Reasoning Improves Acting in LLM Agents. ArXiv:2505.09970
Searle, J. R. 1980. Minds, Brains, and Programs. In Behavioral and Brain Sciences 3 (3): 417–457
Shah, J. D. 2026. Triple-Bottom-Line Sustainability of Language Models for Edge AI: A Comparison between SLMs and Quantized LLMs. ArXiv:2609.00665v1.
Sharma, M., Tong, M., Korbak, T., Duvenaud, D., Bowman, S. R., Perez, E., & Evans, O. 2023. Towards understanding sycophancy in language models. ArXiv:2310.13548.
Xiong, M., et al. 2025. MPO: Boosting LLM Agents with Meta Plan Optimization. ArXiv:2503.02682.
Yao, S., Zhao, J., Yu, D., Du, N., Shafran, I., Narasimhan, K., & Cao, Y. 2022. ReAct: Synergizing Reasoning and Acting in Language Models. ArXiv:2210.03629.
Yilin, G. et al. 2026 Control Illusion: The Failure of Instruction Hierarchies in Large Language Models. The Fortieth AAAI Conference on Artificial Intelligence (AAAI-26).
Wang, G., et al. 2023. Voyager: An Open-Ended Embodied Agent with Large Language Models. ArXiv:2305.16291.
Weber, M. 1913. Über einige Kategorien der verstehenden Soziologie. In Gesammelte Aufsätze zur Wissenschaftslehre. Tübingen: Mohr.
Wright, G. H. von. 1971. Explanation and understanding. Conell, Conell University Press, 1971 .
Wurgaft, D., et al. 2026. Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior. ArXiv:2605.05115v1.
Downloads
Pubblicato
Come citare
Fascicolo
Sezione
Licenza
Copyright (c) 2026 Sabato Danzilli

Questo lavoro è fornito con la licenza Creative Commons Attribuzione 4.0 Internazionale.