Abstrakt
Artykuł przedstawia porównawcze badanie pilotażowe dotyczące tego, jak różne konfiguracje dużych modeli językowych kształtują szkolne narracje historyczne. Wykorzystując stały zestaw pytań (P1–P5) zaczerpnięty z podręcznika Europa. Nasza historia oraz cztery persony AI: domyślny ChatGPT jako baseline (C0), ChatGPT dopromptowany do roli „analitycznego badacza” (C1), NotebookLM z generowaniem wspomaganym wyszukiwaniem (RAG) na wspólnym źródle podręcznikowym (C2) oraz system agentowy pracujący na tej samej bazie wiedzy (C3), analizujemy odpowiedzi jako artefakty dyskursywne o mierzalnych własnościach formalnych i retorycznych. Ramę badania tworzą cztery perspektywy: poprawność merytoryczna i zakotwiczenie w źródle, profil retoryczny i aksjologia, formalna czytelność i „tekstura” leksykalna oraz spójność logiczno-kauzalna. Wskaźniki ilościowe (m.in. długość zdań, polsko-adaptowany wskaźnik czytelności typu Fog, różnorodność słownictwa TTR i MTLD, nominalizacja oraz markery struktury) liczone są na oryginalnych odpowiedziach w języku polskim, bez tłumaczenia. Wyniki pokazują systematyczne kompromisy między zakotwiczeniem a głębokością interpretacji: konfiguracje retrieval-based i agentowe częściej wytwarzają „głos podręcznikowy” i stabilne ramy dydaktyczne, natomiast promptowanie persony wzmacnia metarefleksję i ocenę opartą na kryteriach merytorycznych. Analiza krytycznego przypadku wskazuje ponadto, że pozór ugruntowania może współwystępować z błędem kotwiczącym (hinge-fact failure), gdy pojedyncza błędna informacja historyczna pozostaje osadzona w narracji spójnej lokalnie. Artykuł kończy się wnioskami dla dydaktyki historii, proponując użycie LLM zależne od konfiguracji oraz praktyczne „ścieżki” dla zastosowań szkolnych.
Bibliografia
Cotton D.R.E., Cotton P.A., Shipway J.R., Chatting and cheating: Ensuring academic integrity in the age of ChatGPT, “Assessment & Evaluation in Higher Education” 2024.
Europa. Nasza historia, Wydawnictwa Szkolne i Pedagogiczne - Eduversum, Warszawa 2022.
Gunning R., The Technique of Clear Writing, McGraw-Hill, New York 1952.
Holmes W., Bialik M., Fadel C., Artificial Intelligence in Education: Promises and Implications for Teaching and Learning, Center for Curriculum Redesign, Boston 2019.
Huang L., Yu W., Ma W. et al., A Survey on Hallucination in Large Language Models, “ACM Computing Surveys” 2025.
Imran M., Almusharraf N., Analyzing the role of ChatGPT as a writing assistant at higher education level: A systematic review of the literature, “Contemporary Educational Technology” 2023, 15(4), ep464.
Kasneci E., Sessler K., Küchemann S. et al., ChatGPT for good? On opportunities and challenges of large language models for education, “Learning and Individual Differences” 2023, 103, 102274.
Kincaid J.P., Fishburne R.P., Rogers R.L., Chissom B.S., Derivation of New Readability Formulas (Automated Readability Index, Fog Count and Flesch Reading Ease Formula) for Navy Enlisted Personnel, Research Branch Report 8-75, Naval Technical Training Command, 1975.
Labadze L., Grigolia M., Machaidze L., Role of AI chatbots in education: systematic literature review, “International Journal of Educational Technology in Higher Education” 2023, 20, 56.
Lewis P., Perez E., Piktus A. et al., Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks, “Advances in Neural Information Processing Systems (NeurIPS)” 2020.
McLuhan M., “The Medium is the Message,” in: Durham M. G., Kellner D. M. (eds.), Media and Cultural Studies: KeyWorks, Wiley-Blackwell, Malden-Oxford-Chichester 2012, pp. 100–107.
Miao F., Holmes W. (eds.), Guidance for generative AI in education and research, UNESCO, Paris 2023.
Munaye Y.Y., Admass W., Belayneh Y. et al., ChatGPT in Education: A Systematic Review on Opportunities, Challenges, and Future Directions, “Algorithms” (MDPI) 2025, 18(6), 352.
OECD; Education International, Opportunities, guidelines and guardrails for effective and equitable use of AI in education, OECD Publishing, 2023 (raport/ wytyczne).
OpenAI, GPT-5 System Card, arXiv 2025.
Schick T., Dwivedi-Yu J., Dessì R. et al., Toolformer: Language Models Can Teach Themselves to Use Tools, “Advances in Neural Information Processing Systems (NeurIPS)” 2023.
Sullivan M., Kelly A., McLaughlan P., ChatGPT in higher education: Considerations for academic integrity and student learning, “Journal of Applied Learning & Teaching” 2023, 6(1).
Wang X., Wei J., Schuurmans D. et al., Self-Consistency Improves Chain of Thought Reasoning in Language Models, arXiv 2022.
Wei J., Wang X., Schuurmans D. et al., Chain-of-Thought Prompting Elicits Reasoning in Large Language Models, “Advances in Neural Information Processing Systems (NeurIPS)” 2022.
Werner W., Gralik D., Trzoss A., Media społecznościowe a funkcjonowanie wiedzy historycznej w Polsce. Raport z badañ, „Przegląd Archiwalno-Historyczny” 6 (2019), pp. 211-235.
Vaswani A., Shazeer N., Parmar N., Uszkoreit J., Jones L., Gomez A. N., Kaiser £., Polosukhin I., Attention Is All You Need, “Advances in Neural Information Processing Systems” 30 (2017), pp. 5998-6008
Yao S., Zhao J., Yu D. et al., ReAct: Synergizing Reasoning and Acting in Language Models, “International Conference on Learning Representations (ICLR)” 2023.
Zawacki-Richter O., Marín V.I., Bond M., Gouverneur F., Systematic review of research on artificial intelligence applications in higher education - where are the educators?, “International Journal of Educational Technology in Higher Education” 2019, 16, 39.
Licencja
Prawa autorskie (c) 2026 Wiktor Werner

Utwór dostępny jest na licencji Creative Commons Uznanie autorstwa – Na tych samych warunkach 4.0 Miedzynarodowe.
